Edinburgh Research Explorer Trace-element geochemistry of -hosted inclusions from the Akwatia Mine, West African Craton: implications for diamond paragenesis and geothermobarometry

Citation for published version: De Hoog, C-J, Stachel, T & Harris, JW 2019, 'Trace-element geochemistry of diamond-hosted olivine inclusions from the Akwatia Mine, West African Craton: implications for diamond paragenesis and geothermobarometry', Contributions to Mineralogy and Petrology, vol. 174, 100. https://doi.org/10.1007/s00410-019-1634-y

Digital Object Identifier (DOI): 10.1007/s00410-019-1634-y

Link: Link to publication record in Edinburgh Research Explorer

Document Version: Publisher's PDF, also known as Version of record

Published In: Contributions to Mineralogy and Petrology

Publisher Rights Statement: © The Author(s) 2019

This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

General rights Copyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights.

Take down policy The University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer content complies with UK legislation. If you believe that the public display of this file breaches copyright please contact [email protected] providing details, and we will remove access to the work immediately and investigate your claim.

Download date: 28. Sep. 2021 Contributions to Mineralogy and Petrology (2019) 174:100 https://doi.org/10.1007/s00410-019-1634-y

ORIGINAL PAPER

Trace‑element geochemistry of diamond‑hosted olivine inclusions from the Akwatia Mine, West African Craton: implications for diamond paragenesis and geothermobarometry

J. C. M. De Hoog1 · T. Stachel2 · J. W. Harris3

Received: 8 May 2019 / Accepted: 16 October 2019 / Published online: 14 November 2019 © The Author(s) 2019

Abstract Trace-element concentrations in olivine and coexisting included in from the Akwatia Mine (Ghana, West African Craton) were measured to show that olivine can provide similar information about equilibration temperature, diamond paragenesis and mantle processes as . Trace-element systematics can be used to distinguish harzburgitic from lherzolite ones: if Ca/Al ratios of olivine are below the mantle lherzolite trend (Ca/Al < 2.2), they are derived from a harz- burgitic mantle source, and syngenetic garnets are without exception subcalcic G10 garnets. For harzburgitic olivines that cannot be identifed this way, Na and Ca contents can be used: olivine inclusions with < 60 µg/g Na and Na/Al < 0.7 are all harzburgitic, whereas those with > 300 µg/g Ca or > 60 µg/g Na are lherzolitic. Conventional geothermobarometry indicates that Akwatia diamonds formed and resided close to a 39 mW/m2 conductive geotherm. A similar value can be derived from Al in olivine geothermometry, with TAl-ol ranging from 1020 to 1325 °C. Ni in garnet temperatures is on average somewhat higher (TNi-grt = 1115–1335 °C) and the correlation between the two thermometers is weak, which may be not only due to the large uncertainties in the calibrations, but also due to disequilibrium between inclusions from the same diamond. Calcium in olivine should not be used as a geothermobarometer for harzburgitic olivines, and often gives unrealistic P–T estimates for lherzolitic olivine as well. Diamond-hosted olivine inclusions indicate growth in an extremely depleted (low Ti, Ca, Na, high Cr#) environment with no residual clinopyroxene. They are distinct from olivines from mantle which show higher, more variable Ti contents and lower Cr#. Hence, most olivine inclusions in Akwatia diamonds escaped the referti- lisation processes that have afected most mantle xenoliths. Lherzolitic inclusions are probably the result of refertilisation after undergoing high-degree melting frst. Trivalent cations appear to behave diferently in harzburgitic diamond-hosted olivine inclusions than lherzolitic inclusions and olivine from mantle xenoliths. Some divalent chromium is predicted to be present in most olivine inclusions, which may explain high concentrations up to 0.16 wt% ­Cr2O3 observed in some diamond inclusions. Strong heterogeneity of Cr, V and Al in several inclusions may also result in apparent high Cr contents, and is probably due to late-stage processes during exhumation. However, in general, diamond-hosted olivine inclusions have lower Cr and V than expected compared to mantle xenoliths. Reduced Na activity in depleted limits the uptake of Cr, V and Sc via Na–M3+ exchange. In contrast, Al partitioning in harzburgites is not signifcantly reduced compared to lherzolites, presumably due to uptake of Al in olivine by Al–Al exchange.

Keywords Diamond inclusions · Olivine · Geothermobarometry · Mantle petrology

Communicated by Daniela Rubatto.

Electronic supplementary material The online version of this article (https​://doi.org/10.1007/s0041​0-019-1634-y) contains supplementary material, which is available to authorized users.

* J. C. M. De Hoog 2 Department of Earth and Atmospheric Sciences, University [email protected] of Alberta, Edmonton, AB T6G 2E3, Canada 3 School of Geographical and Earth Sciences, University 1 School of GeoSciences, The University of Edinburgh, Grant of Glasgow, Gregory Building, Glasgow G12 8QQ, UK Institute, James Hutton Road, Edinburgh EH9 3FE, UK

Vol.:(0123456789)1 3 100 Page 2 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

Introduction formation. These conditions can be estimated based on the chemical composition of co-existing minerals within the The subcontinental lithospheric mantle (SCLM) beneath diamond, assuming that they were trapped simultaneously cratons has been studied extensively as it is a key source (i.e., syngenetic) or at least under similar chemical and of information about the geological and tectonic history of P–T conditions. The requirement of equilibrium between the early Earth. As the main source of diamonds, it is also co-existing minerals does not apply to single-element ther- an important economic reservoir. Olivine is a common mometers, such as single cpx thermobarometry (Nimis and inclusion in such diamonds (Sobolev et al. 2008; Stachel Taylor 2000), Al in olivine thermometry (De Hoog et al. and Harris 2008), as well as the most common mineral in 2010) or Ni in garnet (Canil 1996; Ryan et al. 1996), as Earth’s , and may as such be an important tool these are assumed to be in equilibrium with a relatively in elucidating mantle and diamond formation processes. constant source. However, few, if any, studies have evalu- However, due to its low trace element contents and the ated the validity of olivine-based single-element thermom- lack of an empirical framework to interpret olivine data, eters for diamond-hosted inclusions. its full potential has not been explored (De Hoog et al. A further source of uncertainty in interpreting trace ele- 2010; Foley et al. 2013) and it is rarely used as an indica- ment data is the infuence of the chemical environment on tor mineral for diamond prospecting (e.g., Shchukina and trace element partitioning. For example, diferent exchange Shchukin 2018). mechanisms appear to operate for Cr and Al in olivine in Inclusions in diamonds are of particular interest vs. garnet (De Hoog et al. 2010; Witt-Eick- because diamonds are robust containers shielding inclu- schen and O’Neill 2005). As Na may act to charge balance sions from secular processes in the Earth’s mantle, and trivalent cations (De Hoog et al. 2010; Hervig et al. 1986), thus provide a window into early mantle processes that the absence of clinopyroxene in harzburgitic lithologies have been overprinted in mantle suites (Stachel prevalent in the mantle source of diamonds may afect trace and Harris 2008). For example, it has been noted that gar- element distributions and its use in tracing mantle processes net inclusions in diamonds contain a much larger frac- and geothermometry. tion of Ca-depleted garnets than mantle xenoliths from This study aims to evaluate: diamond paragenesis based the same locality, and that olivine inclusions in diamonds on the trace element compositions of olivine inclusions, the can have much higher ­Cr2O3 contents (up to 0.5 wt%) than validity of trace element geothermometry, and the use of commonly recorded in olivine from mantle xenoliths (up olivine to study mantle processes. The diamond suite used is to 0.1 wt%; De Hoog et al. 2010; Hervig et al. 1980b; from Akwatia, Ghana, in which olivine is a common inclu- Sobolev et al. 2009; Stachel and Harris 2008). sion and frequently co-exists with garnet (Stachel and Harris Diamond formation is associated with garnet-bearing sec- 1997a, b). It is, therefore, an ideal suite for the purpose of tions of the SCLM in which harzburgites predominate over this study. lherzolites (Gurney et al. 1993). Harzburgitic assemblages with strongly Ca-depleted garnets (G10 garnets; Dawson and Stephens 1975; Grütter et al. 2004) are of particular inter- Previous work on trace elements in mantle est, as these are indicators for diamondiferous olivine (Gurney 1984; Gurney et al. 1993). Peridotitic inclusions in diamonds have been subdivided into lherzolitic and har- Due to the low concentrations in olivine of many elements zburgitic paragenesis (i.e., mantle source lithologies) based that are major or minor elements in co-existing minerals, on the presence or absence of cpx and the Ca-saturation of trace element data for olivine are relatively scarce. Rou- garnet (Sobolev et al. 1973; Stachel and Harris 2008). How- tine analytical settings for EMPA analysis often result in ever, many diamonds contain olivine inclusions only and inadequate precision and detection limits for elements such currently no criteria exist to date to determine their paragen- as Na, Al, Ca, Cr and Ti in olivine, especially in mantle esis. Attempts to use olivine in diamond exploration have peridotites, where concentrations of these elements are par- mainly focussed on identifying olivine from , i.e., ticularly low. Therefore, analysis of these elements needs to as a kimberlite indicator mineral (e.g., Matveev and Stachel be done by LA-ICP-MS (De Hoog et al. 2010; Foley et al. 2007; Shchukina and Shchukin 2018), but this gives little 2013) or SIMS (Hervig et al. 1986; Steele et al. 1981), or indication about diamond potential. An olivine classifca- using a dedicated trace element routine by EPMA (Batanova tion scheme that would identify harzburgitic olivines would et al. 2015). Alternatively, bulk analysis may yield excellent increase its potential for use as a diamond indicator mineral. results, but extreme care needs to be taken to pick clean Inclusions in diamonds may also provide constraints olivine grains clear of fractures and inclusions (Stead et al. on the pressure–temperature conditions of diamond 2017). Due to their even lower concentrations, most trace elements other than the above-mentioned are only achievable

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 3 of 28 100 by LA-ICP-MS (Stead et al. 2017; Witt-Eickschen and O’Neill 2005), but this is a destructive technique usually not suitable for precious samples such as diamond-hosted inclusions. Despite these analytical challenges, work on trace ele- ments in olivine goes back to the early 1980s. A small set of trace elements in olivine from spinel peridotites was pre- sented by Stosch (1981), who used temperature-dependent partitioning of Sc and Cr between olivine and cpx to derive single-element geothermometers. A set of pioneering stud- ies by Hervig and collaborators (Hervig and Smith 1982; Hervig et al. 1980a, b, 1986; Steele et al. 1981) presented Fig. 1 trace-element data of olivine and co-existing minerals from Photographs of Akwatia diamonds containing silicate mineral inclusions. Left: diamond with red inclusions of garnet; right: a wide range of mafc to ultramafc rocks including kimber- diamond with pale-green to transparent olivine inclusions. Diamond lite-derived lherzolites and harzburgites, as well as olivine size ca. 3 mm inclusion in diamonds. They noticed temperature-dependent partitioning of elements such as Cr, Al and Na, and sug- gested the use of TiO­ 2 to trace mantle metasomatism. A or Ca, whereas the large available dataset of olivine inclu- large excess of Cr was reported in some olivine inclusions sions in diamonds (Sobolev et al. 2009) lacks Na and other in diamonds and a diferent crystallisation environment was trace elements such as V and Sc, whilst Ti is often near the inferred (Hervig et al. 1980b). Temperature dependence of detection limit of the EPMA technique. Due to incomplete trace element partitioning in spinel peridotites was studied datasets, trace element exchange mechanisms in olivine in more detail by Witt-Eickschen and O’Neill (2005), but no from garnet are still poorly constrained, and so Al or Na was measured in their olivines. are potential efects on empirically derived trace element Sobolev et al. (2008, 2009) presented a large dataset of geothermobarometers. This is particularly true for diamond- minor element data from olivine inclusions in diamonds, hosted olivine inclusions. which showed a large variation in elements such as Ca and Cr, and distinct diferences between diamond-hosted olivine inclusions and olivine from mantle xenoliths. De Hoog et al. (2010) presented a large database of olivine trace element Samples data from mostly garnet-bearing lherzolitic mantle xenoliths from a variety of tectonic settings, which allowed them to The alluvial Birim diamond felds near Akwatia, Ghana, calibrate several single-element geothermometers, of which are located within Birimian (Early Proterozoic) metasedi- the Al in olivine thermometer was the most promising, as mentary rocks forming the south-eastern part of the West well as identify geochemical parameters diagnostic for vari- Africa Craton (Chirico et al. 2010; see also Electronic Sup- ous mantle processes, such as melting and refertilisation. A plement, Fig. S1). Akwatia diamonds show little evidence of review paper by Foley et al. (2013) expanded on the data- transport and are associated with metamorphosed ultramafc set by De Hoog et al. (2010) by including a wide suite of rocks with a composition resembling komatiite or boninite- igneous rocks and showing olivine to be a sensitive tracer type volcanic rocks, similar to occurrences at Dachine in of melt metasomatism. Recent work has also provided data French Guyana and at Wawa in Ontario, Canada (Canales for supra-subduction zone xenoliths, an important reservoir and Norman 2003). The eruption age of the diamonds is (Ionov 2010; Pirard et al. 2013; Tollan et al. 2017) miss- around 2.2 Ga (Gurney et al. 2010). They contain abundant ing from previous compilations. The increase in interest of silicate inclusions of peridotitic (Fig. 1), the most abundant the behaviour of H in nominally anhydrous minerals has of which is olivine (66% of inclusions are monomineralic led to many studies about the interaction between trace olivine, another 15% occur in bimineralic and tri-mineralic element uptake in olivine with hydrogen (e.g., Berry et al. assemblages; Stachel and Harris 1997b). The diamonds are 2005, 2007; Tollan et al. 2018), showing that trace element dominantly derived from an initially strongly depleted but partitioning can be strongly infuenced by the chemical subsequently refertilised mantle source (Stachel and Harris environment. 1997a). Despite all the work mentioned above, olivine data from This study presents new major and trace element data for garnet peridotites are still scarce and existing datasets are 28 olivine inclusions from 25 Akwatia diamonds, as well often incomplete. For instance, garnet peridotite data in as for garnets coexisting with 23 of the olivines. Of these Foley et al. (2006) and Glaser et al. (1999) lacked Al and Na inclusions, 5 are of lherzolitic paragenesis (co-existing with

1 3 100 Page 4 of 28 Contributions to Mineralogy and Petrology (2019) 174:100 lherzolitic garnet or cpx), 19 are harzburgitic (co-existing be measured due to their high concentrations in olivine. with harzburgitic garnet) and 4 belong to the peridotitic suite Signifcant molecular isobaric interferences occur for sev- but are of unknown paragenesis (no co-existing garnet or eral elements; most importantly ­MgO+ (impacting mass cpx). 40–42), ­SiO+ (impacting mass 44–46), MgMg­ + (impacting Silicate inclusions were extracted by crushing the host mass 48–52) and MgSi­ + (impacting mass 52–56). The con- diamond, embedded in epoxy and polished to a ¼ µm fn- tribution of these interferences to the isotopes of interest ish. Most inclusions were euhedral to subhedral in shape was determined in the following order: and mostly 100–500 µm in size; typical examples are shown in Fig. 3. No spatial information about the inclusions was 1. 25Mg16O was measured at mass 41 and corrected for recorded, other than that none of the inclusions were 41K (calculated from measured 39K based on their rela- touching. tive isotopic abundance), which allows calculation of 24Mg16O contribution to 40Ca; 2. 28Si16O was measured at mass 44 and corrected for 44Ca (calculated from measured 40Ca based on their relative Analytical methods isotopic abundance), which allows calculation of 29Si16O contribution to 45Sc; Electron microprobe 3. 24Mg25Mg was measured at mass 49 and corrected for 49Ti (calculated from measured 47Ti based on their Major and minor elements were measured by electron probe relative isotopic abundance, after correction of 47Ti for Cameca SX-100 in the School of GeoSciences, University 29Si18O), which allows calculation of 25Mg26Mg contri- of Edinburgh. Major elements Mg, Si and Fe were measured bution to 51V and 26Mg26Mg contribution to 52Cr; using a 10 nA beam current at 15 kV and a counting time of 4. 24Mg29Si + 25Mg28Si was measured at mass 53 and cor- 60 s on both peak and background, whereas minor elements rected for 53Cr (calculated from measured 52Cr based on were measured using a higher beam current of 80 nA and their relative isotopic abundance), which allows calcula- longer counting times of 240 s (Na, Al, Ca, Cr) or 120 s (Mn, tion of 25Mg30Si + 26Mg29Si contribution to 55Mn; Ni) on both peak and background. Calibration standards 5. 29Si30Si contribution to 59Co was estimated using a were forsterite (Mg, Si), fayalite (Fe), wollastonite (Ca), Mg 0.25% production rate of ­SiSi+ from 30Si. spinel (Al), jadeite (Na) and pure metals for the remaining 30 16 62 elements (Cr, Ni, Mn). Accuracy and precision were moni- A small contribution of Si O2– Ni was considered tored by repeat analysis of San Carlos olivine from USGS negligible due to the high Ni content of olivine. Aver- (USNM# 111312/44; Jarosewich et al. 1980), in-house oli- age molecular production rates for ­SiO+/Si+ and ­MgSi+/ vine standard DC0212 (see De Hoog et al. 2010, for prepara- Si+ were 0.29% and 0.11%, respectively. As a check for tion and details) and an in-house Cr- standard, the the validity of the MgMg correction on V and Cr, it was latter mainly to monitor elements which are low in olivine measured on mass 48 as 24Mg24Mg after correction 48Ca, (Na, Al, Ca, Cr). Four or fve analyses were performed for 48Ti and 30Si18O and compared to MgMg measured on each olivine to test their homogeneity. mass 49 (see step 3 above). The diference in V concentra- tion calculated using MgMg measured on mass 48 or 49, respectively, was generally < 3%, showing that the analyti- Ion microprobe—olivine trace elements cal uncertainty added due to the interference corrections was small. Trace elements were measured by secondary ion mass Although the measurements could have been made spectrometry (SIMS) using the Cameca ims-4f at Edin- using high mass resolution of ca. 3000 (M/ΔM) to resolve burgh Ion Microprobe Facility. A 10 nA 16O− primary molecular interferences, this would have been detrimental beam was focussed and impacted onto the samples at for the analysis of elements with low signals, in particular 14.5 kV; sputtered high-energy (75 ± 25 eV) positive Sc, Ti and V. secondary ions were collected at low mass resolution in Calibration was performed using in-house olivine six cycles of 5 s for each isotope, except 47Ti, which was standard DC0212 (LA-ICP-MS value, Table 3 in De Hoog measured for 15 s/cycle. Spot size was approximately et al. 2010), which was measured at least once every 4–8 15 × 20 µm. The following isotopes were measured (those analyses, using 30Si as an internal standard. Standard in brackets were used for interference corrections, see deviation of 15 repeats on the standard was better than 5% details below): 7Li, 23Na, 27Al, 30Si, 31P, 39K, (40Ca), (41K), for all elements except P (10%) and Sc (20%). To verify 43Ca, (44Ca), 45Sc, 47Ti, (48Ti), (49Ti), 51V, 52Cr, (53Cr), that SIMS data gave similar results as LA-ICP-MS data, 55Mn, 59Co, 62Ni. Note that none of the Mg isotopes could olivine grain from selected Kaalvallei xenoliths from De

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 5 of 28 100

Hoog et al. (2010) was re-analysed by SIMS. Agreement is Mineral chemistry overall excellent (Electronic Supplement, Fig. S2). Olivine Ion microprobe—garnet trace elements Major elements Selected trace elements in garnet were measured (Ti, Sc, Ni, V, Co) by SIMS using the Cameca ims-1270 at EIMF, Uni- The olivine inclusions presented here are a subset of a larger 16 − versity of Edinburgh. A O2 primary beam of 1–3 nA was suite of inclusions documented previously by Stachel and impacted onto the samples at 20 kV; sputtered positive ions Harris (1997b). New major element data presented here were collected with no energy fltering in 15 cycles of 4 s (Table 1) agree well with previous data but show less scat- for each isotope. Spot size was approximately 15 × 10 µm. ter due to the improved analytical protocol (Fig. 2a, b). The 45 49 57 50 51 59 Measured isotopes were Sc, Ti, Fe, Cr, V, Co and olivines span a range of forsterite contents from 90.3 to 60 60 30 28 Ni. Potential molecular interferences on Ni ( Si2, SiO2, 94.2% (median 92.9%); most show very little variation in 44 45 29 49 24 25 51 25 26 CaO), Sc ( SiO), Ti ( Mg Mg), V ( Mg Mg) and Fo contents (< 0.1%) with maximum variations up to 0.3% 59 29 30 Co ( Si Si) were avoided using sufciently high mass in only few olivines. The lowest Fo values (90.3–92.9%) 50 resolution (M/ΔM ca. 4000 at 5% peak width). The Cr are from lherzolitic inclusions, whereas harzburgitic olivines 50 50 signal was corrected for small contributions of Ti and V range from 92.6 to 94.3% (average 93.2%). This is identical isobars. to the average mode of 93.2% for harzburgitic olivines from To use Ni in garnet as a thermometer, accuracy of the a global database of olivine inclusions in diamonds (Stachel Ni content is essential. Therefore, in-house garnet stand- and Harris 2008), and slightly higher than the average of ards that were prepared specifcally for this application were 92.9% from a large suite of diamond-hosted inclusions from used for calibration (Ivanic 2007). Nickel contents of the Sobolev et al. (2009), which also included lherzolitic inclu- standards range from 41 to 140 µg/g and they have major sions (Fig. 2b). element contents representative of those observed in mantle- MnO contents range from 0.081 to 0.112 wt% (median derived garnets. The same standards were used to calibrate 0.093 wt%), apart from one high value of 0.15 wt%, and ­TiO2 and Cr­ 2O3. Working curves for Ni, Ti and Cr show show a well-defned negative correlation with Fo contents 2 excellent correlations (r = 0.995–0.999; Electronic Sup- (Fig. 2a), which overlap with the diamond inclusions from 57 plement, Fig. S3). Fe was used as internal standard using Sobolev et al. (2009). NiO contents show a narrow range FeO contents measured by electron probe micro-analysis. from 0.36 to 0.43 wt% (median 0.404 wt%), which is higher Sc and Co were calibrated using two garnet standards (KP1, than the global array of NiO contents in olivine inclusions in Binge Binge Point, and DDI, Dutsen Dushowo; Irving and diamonds (0.36–0.38 wt%; Stachel and Harris 2008; Sobolev Frey 1978). As V content is not known for any of the garnet et al. 2009). There is no correlation of NiO with Fo content, standards in this study, it was measured by SIMS Cameca although the spread in NiO contents is larger for the olivines 16 − ims-4f at EIMF, University of Edinburgh. A O beam of with the highest Fo contents (Fig. 2b). Lherzolitic olivines 5 nA was impacted onto the samples at 14.5 kV; sputtered are near the high end of the range (0.40–0.43 wt%), with positive ions were collected with 75 V energy ofset in 10 the exception of anomalous lherzolitic olivine (G201–202), cycles of 5 s for each isotope. Spot size was approximately which also has low Fo and high MnO. 30 × 25 µm. Measured isotopes were 30Si, 45Sc, 47Ti, 51V, 52Cr, 55Mn and 60Ni. Molecular interferences were negated Trace elements using high mass resolution (M/ΔM ca. 2400 at 5% peak width). Basaltic glasses GSE1-G and GSD1-G were used Trace element contents of olivine inclusions are presented for calibration (Jochum et al. 2005). Vanadium contents of in Table 2. Lithium contents show a relatively wide range KP1 and DDI garnet standards were determined as 41 and of concentrations from 0.6 to 2.8 µg/g (median 1.1 µg/g; 134 µg/g, respectively. Other elements agreed well with the harzburgitic olivines are < 1.6 µg/g), which spans the whole data by Irving and Frey (1978). range of olivines from garnet-bearing peridotites from De Hoog et al. (2010).1 Lithium contents do not show any cor- relation with other elements. Sodium contents show a large

1 Note that the olivine data from De Hoog et al. (2010) include data from garnet-free spinel peridotites and other rock types. When refer- ring to these olivine data, we only included garnet-bearing peridotite xenoliths, unless otherwise stated.

1 3 100 Page 6 of 28 Contributions to Mineralogy and Petrology (2019) 174:100 92.74 (12) 3.002 0.008 0.001 0.001 1.846 0.002 0.144 0.001 0.001 0.049 (6) 0.998 51.26 (7) 2gt,2ol 100.03 0.088 (6) l (high-Ti) 0.407 (9) 7.15 (11) G113–135 0.017 (4) 41.30 (23) 0.022 (2) 0.050 (5) 92.88 (10) 3.000 0.008 0.001 0.001 1.846 0.002 0.142 0.001 0.001 0.054 (16) 0.999 51.34 (41) 2gt,2ol 100.00 0.090 (4) l (high-Ti) 0.401 (12) 7.02 (7) G113–134 0.018 (6) 41.42 (21) 0.024 (2) 0.050 (10) 93.40 (13) 2.998 0.008 0.000 0.001 1.853 0.002 0.131 0.001 0.000 0.024 (1) 1.002 51.57 (56) 2gt,3ol 99.93 0.092 (3) h 0.388 (9) 6.50 (7) G112–129 0.010 (4) 41.56 (32) 0.015 (1) 0.031 (8) 93.64 (9) 3.007 0.008 0.000 0.000 1.876 0.002 0.127 0.001 0.001 0.016 (1) 0.992 52.79 (71) gt,ol 101.85 0.082 (4) h 0.414 (9) 6.39 (9) G106–113 0.007 (1) 41.62 (44) 0.025 (10) 0.038 (10) 92.62 (8) 2.993 0.008 0.000 0.000 1.828 0.002 0.146 0.001 0.001 0.018 (2) 1.007 51.20 (30) gt,2ol 100.78 0.099 (9) h 0.412 (5) 7.27 (6) G105–111 0.007 (1) 42.03 (16) 0.032 (7) 0.033 (10) 92.59 (13) 2.991 0.008 0.000 0.000 1.825 0.002 0.146 0.000 0.000 0.015 (2) 1.008 50.73 (42) 2gt,2ol 99.99 0.095 (1) h 0.411 (7) 7.24 (13) G103–106 0.004 (6) 41.78 (55) 0.011 (1) 0.025 (3) 92.63 (12) 3.000 0.008 0.000 0.001 1.842 0.002 0.147 0.001 0.000 0.033 (5) 1.000 51.35 (62) 2gt,2ol 100.46 0.103 (7) h 0.403 (6) 7.29 (12) G103–105 0.005 (4) 41.55 (43) 0.013 (4) 0.031 (4) 91.51 (22) 3.003 0.008 0.000 0.001 1.823 0.002 0.169 0.001 0.001 0.0572 (8) 0.997 50.77 (150) 2gt,2cpx,2ol 101.72 0.112 (4) l 0.424 (10) 8.40 (5) G50–84 0.009 (2) 41.38 (113) 0.022 (4) 0.039 (3) 91.50 (11) 3.002 0.008 0.001 0.001 1.822 0.002 0.169 0.001 0.000 0.055 (1) 0.997 50.41 (41) 2gt,2cpx,2ol 100.16 0.111 (5) l 0.430 (8) 8.35 (7) G50–80 0.013 (7) 41.14 (80) 0.012 (3) 0.038 (4) 94.25 (10) 2.995 0.007 0.001 0.000 1.865 0.002 0.114 0.002 0.001 0.018 (2) 1.003 52.40 (71) gt,ol 101.22 0.081 (2) h 0.383 (8) 5.70 (9) G40–68 0.012 (6) 42.01 (23) 0.040 (5) 0.125 (32) 92.95 (11) 3.004 0.008 0.001 0.001 1.856 0.002 0.141 0.001 0.001 0.024 (3) 0.995 51.56 (12) gt,2ol 100.82 0.093 (5) h 0.404 (6) 6.97 (12) G34–54 0.013 (6) 41.22 (47) 0.023 (4) 0.032 (5) 93.10 (27) 2.991 0.008 0.000 0.000 1.835 0.002 0.136 0.000 0.000 0.0136 (5) 1.009 51.67 (143) gt,ol 101.88 0.088 (3) h 0.413 (6) 6.83 (11) G29–41 0.002 (1) 42.37 (41) 0.011 (5) 0.012 (3) 92.27 (8) 2.997 0.008 0.000 0.001 1.827 0.002 0.153 0.001 0.001 0.056 (2) 1.003 51.39 (51) ol,2cpx 102.28 0.103 (3) l 0.418 (4) 7.68 (2) G23–31 0.010 (5) 42.05 (33) 0.020 (7) 0.051 (12) 90.19 (9) 3.001 0.007 0.000 0.002 1.793 0.003 0.195 0.000 0.001 0.095 (1) 0.999 49.43 (15) 100.74 0.138 (5) 0.380 (7) 9.59 (9) SC-ol 0.007 (2) 41.05 (16) 0.040 (2) 0.015 (2) Compositions of olivine by EPMA of olivine by Compositions 3 3 2 O O O 2 2 2 Fo Sum #Ni #Na #Ca #Mg #Mn #Fe2+ #Cr Na #Al CaO SiO #Si MgO Assemblage Total MnO Paragenesis NiO Al Cr FeO 1 Table Sample

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 7 of 28 100 3.014 0.007 0.001 0.001 1.885 0.002 0.132 0.001 0.001 0.027 0.985 0.086 0.366 6.62 0.016 0.029 0.054 93.47 53.20 41.45 102.32 gt,2ol h G31–44 93.18 (11) 2.993 0.008 0.001 0.000 1.840 0.002 0.135 0.001 0.001 0.019 (1) 1.006 51.41 (34) gt,ol,chr 101.12 0.090 (6) h 0.403 (5) 6.71 (7) G304–307 0.012 (2) 41.93 (20) 0.023 (3) 0.0478 (6) 93.51 (10) 3.011 0.008 0.000 0.000 1.880 0.002 0.130 0.000 0.000 0.008 (1) 0.989 52.41 (41) gt,ol 100.12 0.088 (4) deep-h 0.424 (2) 6.48 (6) G303–304 0.003 (3) 41.11 (42) 0.0031 (7) 0.022 (1) 93.54 (22) 2.999 0.007 0.000 0.001 1.858 0.002 0.128 0.002 0.001 0.0236 (7) 1.000 51.76 (97) gt,ol 100.76 0.093 (3) h 0.360 (9) 6.37 (13) G302–303 0.008 (4) 41.53 (43) 0.028 (4) 0.113 (19) 92.68 (14) 2.985 0.008 0.000 0.001 1.816 0.002 0.143 0.001 0.000 0.020 (2) 1.014 50.82 (104) gt,ol,opx 101.35 0.097 (5) h 0.409 (7) 7.15 (3) G209–217 0.005 (2) 42.31 (54) 0.010 (3) 0.052 (4) 93.52 (12) 3.010 0.007 0.000 0.001 1.879 0.002 0.130 0.001 0.000 0.027 (3) 0.990 53.20 (57) gt,ol 102.55 0.090 (4) h 0.383 (13) 6.57 (7) G208–213 0.004 (2) 41.77 (51) 0.011 (2) 0.037 (3) 93.44 (17) 2.998 0.008 0.000 0.000 1.856 0.002 0.130 0.000 0.000 0.007 (3) 1.001 51.17 (65) gt,ol 98.85 0.089 (4) h 0.411 (10) 6.41 (17) G202–205 0.005 (6) 41.16 (55) 0.016 (5) 0.020 (12) 90.32 (32) 3.015 0.007 0.000 0.002 1.820 0.003 0.195 0.001 0.001 0.086 (10) 0.984 49.84 (72) gt,ol,cpx 99.88 0.149 (3) l (high-Ti) 0.361 (5) 9.52 (23) G201–202 0.009 (3) 40.18 (50) 0.036 (5) 0.048 (2) 93.78 (29) 2.985 0.007 0.000 0.001 1.838 0.002 0.122 0.001 0.001 0.0204 (6) 1.014 51.57 (69) ol,chr 100.20 0.088 (4) p 0.384 (15) 6.10 (23) G127–160 0.005 (2) 42.42 (32) 0.019 (1) 0.027 (2) 93.08 (18) 2.999 0.008 0.000 0.001 1.849 0.002 0.137 0.001 0.001 0.027 (2) 1.001 51.50 (31) 2gt,2ol,opx 100.06 0.097 (4) h 0.399 (6) 6.82 (16) G126–158 < 0.002 41.56 (31) 0.018 (4) 0.041 (2) 92.74 (8) 2.996 0.008 0.000 0.001 1.837 0.002 0.144 0.001 0.001 0.0289 (4) 1.002 50.81 (18) 2gt,2ol,opx 99.57 0.096 (5) h 0.412 (6) 7.09 (6) G126–156 0.002 (2) 41.33 (19) 0.029 (3) 0.078 (6) p peridotitic 93.29 (10) 3.005 0.008 0.000 0.001 1.863 0.002 0.134 0.001 0.000 0.024 (3) 0.995 52.28 (33) ol 100.80 0.096 (9) p 0.413 (11) 6.71 (7) G120–147 0.006 (1) 41.60 (21) 0.015 (5) 0.061 (18) 92.76 (7) 2.997 0.008 0.000 0.001 1.840 0.002 0.144 0.001 0.000 0.023 (9) 1.002 51.30 (25) ol,opx 100.43 0.097 (4) p 0.427 (5) 7.14 (6) G119–145 0.003 (2) 41.66 (6) 0.015 (3) 0.0325 (3) harzburgitic, h harzburgitic, 93.25 (5) 3.004 0.007 0.000 0.000 1.862 0.002 0.135 0.001 0.000 0.016 (1) 0.996 51.88 (11) 2ol 100.17 0.095 (2) p 0.373 (5) 6.70 (4) G118–143 0.003 (1) 41.37 (10) 0.011 (2) 0.045 (10) (continued) 3 3 2 O O O 2 2 2 Fo Sum #Ni #Na #Ca #Mg #Mn #Fe2+ #Cr Na #Al CaO SiO #Si MgO Assem - blage Total MnO Paragenesis NiO Al Cr FeO Sample G31–44 which for parentheses, except as the indicated between in thedigit(s) and are variation on the last same grain expressed are based on 4–5 analyses deviations Oxides in wt%. Standard et al. 1980 ) run based on sessions. Cation formula olivine USNM #111312/44 (Jarosewich on San Carlos throughout the analytical of 34 analyses once. SC-ol is the average only analysed was the whereas the second number represents number of found, that the ‘G’ in the number behind the name represents 4O. Note sample letter number of theinclusion was diamond in which the inclusion l lherzolitic, Paragenesis: 1 Table

1 3 100 Page 8 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

0.20 0.50 ABlherzolitic peridotitic 0.16 harzburgitic 0.45

0.12 0.40 O (wt.% ) O (wt.% )

0.08 Ni 0.35 Mn

0.04 0.30

0.00 0.00 89.090.0 91.092.0 93.0 94.095.089.090.0 91.092.093.0 94.095.0 Fo content Fo content

Fig. 2 a MnO and b NiO (wt%) content of olivine inclusions in tia diamonds analysed at lower precision (Stachel and Harris 1997b), Akwatia diamonds vs. their forsterite (Fo%) content. Error bars (1s) whereas small plus signs represent diamond-hosted olivine inclusions based on 4–5 repeat analyses on diferent locations on each grain. from Sobolev et al. (2009) Grey dots are data from a larger suite of olivine inclusions in Akwa- range (1.3–115 µg/g, median 28 µg/g), but are low com- from 20 to 620 µg/g (De Hoog et al. 2010), as well as olivine pared to lherzolitic olivines from garnet peridotite xeno- inclusions in diamonds (Sobolev et al. 2009, which mostly liths (up to 300 µg/g; De Hoog et al. 2010). Harzburgitic range from 30 to 350 µg/g, although a few higher values do olivines have < 57 µg/g Na. Aluminium contents also show occur. Cr# (#Cr/[#Cr + #Al]) range from 0.47 to 0.90, with a wide range from 13 to 116 µg/g (median 64 µg/g), with an average of 0.72, which are also high compared to mantle harzburgitic and lherzolitic olivine inclusions being similar xenoliths (0.33–0.75, average 0.64) and diamond inclusions and showing a comparable range to olivines from mantle (average 0.54). Cobalt values range from 116 to 164 µg/g xenoliths (De Hoog et al. 2010) and olivine inclusions in (median 133 µg/g), with no diference between harzburgitic diamonds (Sobolev et al. 2009). Calcium concentrations and lherzolitic inclusions, and are very similar to olivine fall between 34 and 565 µg/g (median 148 µg/g); the high- inclusions in diamonds elsewhere (100–181 µg/g, average est values are from lherzolitic olivines. Scandium contents 129 µg/g; Sobolev et al. 2009). range from 0.5 to 2.6 µg/g (median 1.0 µg/g) and show no Duplicate ion microprobe analyses of olivine generally correlation with other elements. Titanium contents are low show excellent reproducibility for most elements (e.g., Mn, (< 72 µg/g, median 2.2 µg/g) with values > 10 µg/g all from Ni, Co, Ti) but occasionally large diferences for Cr and lherzolitic inclusions. The latter co-exist with Ti-rich gar- V, and, to a lesser extent, Na, Al and Ca. This variability nets, but still have considerable less Ti than Ti-rich olivines was confrmed by high-resolution line scans across several in lherzolitic mantle xenoliths (up to 270 µg/g; De Hoog olivines by EPMA (Fig. 3) as well as a Cr map of one of the et al. 2010). The highest Ti content in harzburgitic olivine grains (Electronic Supplement, Fig. S4). The patterns are is 9.6 µg/g, but the median is only 1.0 µg/g. mostly highly irregular for Cr but fat for Mn, whereas Ca Vanadium contents vary from 0.7 to 10.3 µg/g (median and Al correlate with Cr for some inclusions (G106–113, 4.8 µg/g), with one exception of 15 µg/g. This is similar to G112–129; Fig. 3) and are fat in others (G105–111; Fig. 3). the range observed in olivine from the xenoliths (1–10 µg/g; The variability in Cr and V may be due to the presence of De Hoog et al. 2010). A weak positive correlation of V can nano-inclusions of chromite, which have been reported in a be observed with Cr, but this correlation is considerably less olivine inclusion in diamond from Yakutia (Sobolev et al. clear than in olivines from mantle xenolith (De Hoog et al. 2008), but no such inclusions could be observed in Akwatia 2010). Chromium contents range from 120 to 1100 µg/g olivines even at very high magnifcation using BSE imag- (median 350 µg/g) with little diference between harzbur- ing. Hence, if such inclusions exist they must be smaller gitic and lherzolitic inclusions. These values are consider- than 0.1 µm, in which case their presence cannot explain the ably higher than those from mantle xenoliths, which range large variability under the ion microprobe beam (which is

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 9 of 28 100

Table 2 Trace-element Sample Par. Fo Li Na Al Ca Sc Ti V Cr Mn Co concentrations of olivine inclusions by SIMS G23–31 l 92.3 0.9 62 80 351 0.7 5.8 5.1 350 797 142 G29–41 h 93.1 1.1 3 38 84 0.7 0.2 2.5 130 697 133 G30–40 h 92.3 0.9 10 29 117 0.7 4.5 4.7 243 725 125 G32–46 h 92.6 1.3 11 77 215 1.1 2.7 3.5 320 756 134 G32–46 h 92.6 1.2 5 72 175 1.5 2.3 3.5 280 777 134 G34–54 h 92.9 0.9 43 38 133 0.6 0.3 1.6 191 706 128 G37–60 h 93.0 1.0 6 58 67 1.0 1.5 3.7 196 702 119 G38–63 h 93.3 0.9 1 69 64 1.0 0.2 3.4 125 685 115 G40–68 h 94.2 0.6 33 63 70 1.0 3.7 6.7 791 597 121 G50–84 l 91.5 1.0 71 39 376 1.1 11.6 5.8 284 929 163 G103–105 h 92.6 0.8 57 56 233 0.8 0.6 1.7 234 778 138 G103–106 h 92.6 1.3 27 53 173 0.9 0.6 1.5 165 703 129 G105–111 h 92.6 1.6 1 65 98 1.4 1.1 4.9 225 799 140 G106–113 h 93.6 1.0 31 92 93 1.5 9.6 3.5 419 646 126 G112–129 h 93.4 1.3 28 102 151 1.9 1.9 7.1 638 696 126 G112–129 h 93.4 1.1 21 70 147 1.8 2.3 2.4 337 711 127 G113–134 l-Ti 92.9 1.2 115 108 383 1.3 33.1 8.6 446 662 127 G113–135 l-Ti 92.7 1.1 112 116 370 2.1 34.8 9.0 443 709 139 G113–135 l-Ti 92.7 0.9 115 123 416 1.4 33.8 9.3 450 717 135 G118–143 p 93.2 1.1 73 109 138 1.4 19.6 6.2 1043 698 122 G118–143 p 93.2 1.3 34 67 109 1.4 17.9 3.1 500 702 121 G119–145 p 92.8 1.3 33 89 256 0.9 2.1 6.8 499 827 158 G119–145 p 92.8 1.0 34 86 232 1.2 2.1 4.8 439 757 136 G120–147 p + 93.3 + 1.9 + 83 + 90 + 146 + 0.9 + 5.2 + 15.5 + 1097 + 748 + 145 G120–147 p 93.3 1.8 30 61 146 1.4 5.4 2.8 382 710 131 G126–156 h 92.7 1.2 18 92 177 1.1 0.6 7.5 562 810 148 G126–158 h 93.1 1.2 16 75 167 1.3 0.5 4.6 281 739 132 G126–158 h 93.1 1.1 11 75 171 1.4 0.6 5.1 290 728 131 G127–160 p 93.8 1.0 15 67 125 1.6 2.6 1.2 156 698 125 G201–202 l-Ti 90.3 2.8 94 94 563 2.6 71.7 10.0 399 1166 164 G202–205 h 93.4 1.4 4 13 34 0.5 0.2 0.7 118 721 146 G208–213 h 93.5 0.9 29 54 196 0.8 1.2 5.1 237 704 134 G209–217 h 92.7 1.4 27 47 133 0.8 0.9 5.1 361 710 128 G302–303 h 93.5 1.4 37 61 136 1.2 3.9 2.5 852 705 133 G303–304 h 93.5 0.9 16 23 63 0.7 0.5 3.5 380 659 123 G303–304 h 93.5 1.1 12 32 59 0.6 0.7 1.8 136 659 120

All concentrations in µg/g. Fo = forsterite content of olivine (data for G30–40, G32–46, G37–60, G38–63 from Stachel and Harris 1997b). Inclusions with the same name as the row above indicate repeat analyses of the same grain. Inclusions with the same number in the sample name following the letter G are from the same diamond. The analysis marked with plus signs (G120–147) overlapped a healed microfracture par. paragenesis: l lherzolitic, h harzburgitic, p peridotitic, l-Ti Ti-rich lherzolite ca. 10 × 15 µm in diameter, i.e., many orders of magnitude as small bright spots in BSE imaging, presumably micro- larger) unless they are distributed highly heterogeneously. inclusions of chromite. A Cr concentration map of inclusion G106–113 (Electronic Supplement; Fig. S4) does show considerable heterogeneity Garnet on a scale of 2–10 µm, which is consistent with the variabil- ity in EPMA and SIMS Cr data. The origin of this heteroge- Major element compositions of garnets in this study were neity will be discussed later. presented by Stachel and Harris (1997b) and are summarised Healed micro-fractures in two olivine grains are distinctly here. The garnets show variable but high Cr­ 2O3 contents diferent and show high contents of Cr, Ca and Al as well (4.3–20.6 wt%, Cr# 0.13–0.66) and are some of the most

1 3 100 Page 10 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

1.000 G40-68 A B

0.100 Mn

B Cr

wt.% oxid e Ca 0.010 Al A

0.001 050100 150 µm

1.000 G106-113 A B CD A B 0.100 Mn C D Cr

wt.% oxid e Ca 0.010 Al

0.001 050100 µm 050100 µm

1.000 G105-111 A B B

0.100 Mn

Cr

A wt.% oxid e Ca 0.010 Al

0.001 050100 µm

G112-129 1.000 fractures A A B

0.100 Mn

Cr Ca wt.% oxid e 0.010 Al fractures

B

0.001 050100 150200 µm

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 11 of 28 100

◂Fig. 3 Concentration profles (on a log scale) by EPMA of Mn, Cr, within the diamond stability feld, as required for inclusions Ca and Al in selected olivine inclusions. The analytical uncertainty in diamonds. We will now compare the results from various of each point is generally equal or smaller than the symbol size. Location of profles are indicated in BSE images on the left, which geothermobarometers in detail. also show the locations of ion microprobe pits. See also Fig. S4 in Simultaneous P–T information could be obtained the electronic supplement for a high-precision Cr map of inclusion from seven single cpx inclusions (present in four of the G106–113 studied diamonds; mineral data from Stachel and Harris 1997b). Using the single clinopyroxene geothermobarom- Cr-rich garnets reported in the literature. Mg# range from eter from Nimis and Taylor (2000), we obtained values of Tcpx Pcpx 83.6 to 89.1%. CaO contents are variable (1.0–6.3 wt %) and NT00 = 1185–1200 °C and NT00 = 53–54 kbar for three 27 out of 32 garnets are subcalcic, i.e., these are harzburgitic (in two diferent diamonds, G50 and G201) and or G10 garnets (Table 3). MnO contents span a relatively 1300–1310 °C at 59–60 kbar for two others (in a single narrow range (0.22–0.35 wt%) and show a positive correla- diamond, G23). Two anomalously K-rich cpx (0.7 wt% tion with FeO. ­K2O) in a single diamond (G41) with no other inclusions Tcpx Newly obtained trace element data (Table 3) show that gave relatively low NT00 of 1150–1235 °C at 58 kbar, but Ti contents of garnets span an extremely wide range from the high K­ 2O contents are outside the calibration range of 4 to 4600 µg/g, although harzburgitic garnets all have Ti Nimis and Taylor (2000). Overall, the single-cpx results contents < 470 µg/g (< 0.078 wt% ­TiO2). Scandium and V are about 50 °C cooler than the Kalahari geotherm of contents are high (130–560 µg/g and 250–505 µg/g, respec- Rudnick and Nyblade (1999) and fall close to a 39 mW/ tively). Cobalt (40–48 µg/g) shows very little variation. A m2 model conductive geotherm (Hasterok and Chapman weak negative correlation exists between V and Mg#, as 2011) (Fig. 4). well as a weak positive correlation between Sc and Cr, but For cpx co-existing with garnet (in diamonds G50 and Tgrt-cpx no other elements show any correlation, including V and Cr. G201), temperature estimates ( Kr88 ) based on garnet–cpx Nickel contents range from 75 to 141 µg/g. Mg–Fe exchange thermometry (Krogh 1988) are about Tcpx 35 °C cooler than NT00 for both diamonds using the same Pcpx NT00 barometer (i.e., agree within error, which are ± 25 °C Tgrt-cpx Tcpx Geothermobarometry and ± 30 °C for Kr88 and NT00 , respectively). Two diamonds (G126 and G209) contained opx in addi- Conventional major‑element geothermobarometry tion to garnet and olivine; therefore, equilibration pres- sures could be calculated using Al–opx thermobarometry Equilibrium pressures and temperatures of diamond-hosted (Brey and Köhler 1990) combined with garnet–opx (Har- inclusions at the time of their entrapment can be estimated ley 1984) or garnet–olivine (O’Neill and Wood 1979) based on the chemical composition of co-existing miner- Mg–Fe exchange thermometers (Fig. 4). For diamond als in the same diamond, assuming that they were trapped G126, these thermometers give a wide range of tempera- Tgrt-ol Tgrt-opx simultaneously or at least under similar chemical and P–T tures ( OW79 = 1150–1280 °C, Ha84 = 1160–1185 °C), conditions, and that they are not touching other inclusions, in which renders the pressure estimate rather uncertain PAl-opx which case they would record the temperature of the mantle ( BKN = 50–61 kbar). As the two olivine inclusions in this at the time of eruption (Phillips et al. 2004). All Akwatia diamond have quite diferent compositions (Table 2), one inclusions reported here were non-touching, and for most or both of these are not in equilibrium with co-existing opx Tgrt-ol diamonds it was possible to calculate temperature estimates, and garnet. Indeed, if we ignore olivine G126–156, OW79 Tgrt-opx but pressure estimates could only be derived for the few results (1150–1180 °C) are much closer to Ha84 and PAl-opx Tgrt-ol Tgrt-opx diamonds that contained garnet–opx pairs or cpx, the only BKN = 50–53 kbar. For diamond G209, OW79 and Ha84 assemblages that can be reliably used as geobarometers. A difer by 55 °C (i.e., agree within error due to the larger summary of P–T estimates is given in Table 4. Note that uncertainties of these geothermometers of 60 °C and 40 °C, PAl-opx equilibrium temperatures have been rounded to the nearest respectively), resulting in a BKN estimate of 50–55 kbar. fve degrees to refect uncertainty in geothermometer cali- Two more diamonds contained garnet and opx but no brations and chemical analyses upon which the calculations olivine. Diamond G104 contained single garnet and opx PAl-opx Tgrt-opx are based. Most geothermobarometer calibrations have an inclusions giving a BKN at Ha84 estimate of 1210 °C accuracy of 15–30 °C and 2–3 kbar (e.g., Brey and Köhler at 58 kbar. Diamond G111 contained two garnets and 1990; Krogh 1988; Nimis and Taylor 2000). The estimated three opx, which resulted in an wide range of P–T esti- pressures and temperatures span the whole range of condi- mates (1160–1450 °C and 43–78 kbar) due to the range in 2 tions expected for diamonds sampled along a 38–40 mW/m ­Al2O3 contents in three co-existing opx (0.92–1.19 wt%). conductive geotherm (Hasterok and Chapman 2011), i.e., As the Mg# of the two garnet inclusions was also sig- from 1100 °C at 45 kbar to 1300 °C at 65 kbar, and all fall nifcantly diferent (Mg# 89.9 vs. 90.3), inclusions in this

1 3 100 Page 12 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

Table 3 Trace-element Garnet Mg# Cr# Ca_int Grt type Ti Ni Sc V Co concentrations of garnet inclusions by SIMS G29–39 0.88 0.26 0.86 G10 52 119 166 335 47 G30–42 0.85 0.29 1.99 G10 243 104 161 426 47 G31–45 0.88 0.28 0.79 G10 65 141 174 378 47 G32–48 0.88 0.34 1.13 G10 115 116 251 293 45 G33–50 0.87 0.16 1.73 G10 4 79 157 346 44 G33–51 0.87 0.15 1.71 G10 4 81 167 332 44 G34–53 0.87 0.29 1.96 G10 19 90 227 308 44 G35–57 0.87 0.16 1.72 G10 4 82 174 345 45 G37–61 0.87 0.44 0.64 G10 87 103 423 309 47 G38–65 0.87 0.30 0.68 G10 7 88 272 305 45 G40–67 0.86 0.40 0.96 G10 n.a. n.a. n.a. n.a. n.a. G50–82 0.84 0.20 4.50 G9 1325 90 216 394 45 G50–85 0.83 0.21 4.23 G9 1430 91 203 433 45 G103–103 0.87 0.16 2.05 G10 21 94 156 312 44 G103–104 0.88 0.16 2.03 G10 22 92 156 344 41 G104–107 0.88 0.18 3.14 G10 n.a. n.a. n.a. n.a. n.a. G105–109 0.87 0.34 0.84 G10 92 98 248 342 45 G106–112 0.89 0.24 0.62 G10 450 130 141 326 46 G107–114 0.88 0.31 0.73 G10 61 99 377 149 44 G111–122 0.88 0.23 0.45 G10 n.a. n.a. n.a. n.a. n.a. G111–123 0.88 0.25 0.46 G10 n.a. n.a. n.a. n.a. n.a. G112–127 0.89 0.31 0.91 G10 94 124 199 292 44 G112–128 0.89 0.30 0.92 G10 101 123 214 313 44 G113–132 0.87 0.16 3.80 G11 2259 98 137 330 41 G113–133 0.87 0.15 3.87 G11 2182 97 130 320 41 G126–154 0.88 0.29 1.39 G10 29 104 232 352 44 G126–155 0.87 0.29 1.37 G10 29 108 232 337 45 G201–203 0.84 0.13 4.47 G11 4583 125 256 475 44 G202–204 0.86 0.46 0.70 G10 31 92 217 506 46 G208–214 0.88 0.35 2.14 G10 280 90 241 338 40 G209–216 0.86 0.32 1.53 G10 n.a. n.a. n.a. n.a. n.a. G302–302 0.88 0.51 0.95 G10 242 123 236 284 44 G303–305 0.86 0.66 0.62 G10 28 128 234 375 45 G304–308 0.87 0.35 0.97 G10 214 87 383 248 45 G305–309 0.88 0.22 1.26 G10 464 105 186 331 45 G306–311 0.88 0.24 0.50 G10 111 100 167 271 47 G307–313 0.86 0.32 1.87 G10 45 75 330 349 44

Trace elements by SIMS in µg/g, other chemical data based on EPMA (previously published in Stachel and Harris 1997b): Mg# = atomic Mg/(Mg + Fe), Cr# = atomic Cr/(Cr + Al), Ca_int (CaO intercept value) indi- cates how far a garnet is removed from the –lherzolite division (Ca_int = 3.375) after Grütter et al. (2004). Garnet types: G10 harzburgitic, G9 lherzolitic, G11 high-Ti lherzolitic (Grütter et al. 2004). Inclusions with the same number in the sample name following the letter G are from the same diamond n.a. not analyzed particular diamond are clearly not syngenetic, but were garnet and olivine inclusions range from 975 to 1440 °C trapped at diferent times in a chemically/physically evolv- at a fxed pressure of 55 kbar, and most cluster around ing environment. 1100–1300 °C (average 1210 °C). In general, if multiple For all other diamonds, only temperatures could be esti- garnets and/or olivines were present within a diamond, they mated due to the lack of cpx and opx. The most common fall within 50 °C with only a few exceptions. An updated cal- Tgrt-ol co-existing mineral pair is garnet–olivine. OW79 results for ibration of the garnet–olivine Fe–Mg exchange thermometer

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 13 of 28 100

Table 4 Pressure–temperature estimates cpx grt-cpx grt-opx Thermometer T T T Tgrt-ol TAl-ol TNi-grt NT00 Kr88 Ha84 OW79 DH10 Can99 cpx cpx Barometer P P PAl-opx PAl-opx P* NT00 NT00 BKN BKN Uncertainty (kbar/°C) ± 2/± 30 ± 2/± 25 ± 2/± 40 ± 2/± 60 ± 2/± 20 ± 50

Diamond Inclusions kbar °C kbar °C kbar °C kbar °C kbar °C °C

G23 ol,2cpx 60 1305 *57 1240 59 1295 *56 1240 G29 gt,ol 46 1050 1230 G30 gt,ol 47 1065 1335 G32 gt,2ol 57 1225 1260 1235 G34 gt,2ol 49 1095 1165 1165 G37 gt,ol,chr 52 1140 1185 G38 gt,2ol,chr 49 1095 1180 1175 G40 gt,ol 58 1245 n.a. G41 2cpx 58 1150 58 1235 G50 2gt,2cpx,2ol 53 1200 51 1160 *49 1105 1150 50 1120 1155 54 1185 55 1200 *60 1330 1155 53 1160 1160 G103 2gt,2ol 53 1155 1180 1170 51 1130 1175 1170 G104 gt,opx 58 1210 G105 gt,2ol 54 1180 1185 1190 G106 gt,ol 60 1270 1255 G111 2gt,3opx 75 1465 51 1245 78 1450 65 1355 43 1160 68 1340 G112 2gt,3ol 62 1305 1255 1270 1240 G113 2gt,2ol 62 1310 1190 1185 63 1325 1185 1180 G126 2gt,2ol,opx 61 1280 + 64 1320 1200 58 1250 + 65 1320 1205 53 1185 52 1180 + 55 1210 1205 51 1160 50 1150 + 55 1210 1215 G201 gt,ol,cpx 53 1185 52 1150 *56 1260 1280 G202 gt,ol 39 920 1165 G208 gt,ol 53 1160 1180 G209 gt,ol,opx 50 1095 55 1150 + 57 1170 n.a. G302 gt,ol 57 1235 1275 G303 gt,ol 45 1025 1240

1 3 100 Page 14 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

Table 4 (continued)

Temperatures rounded to the nearest 5 °C. P*: TAl-ol with either PNT00 (marked *) or PBKN for pressure estimate (marked +), or transposed onto 39 mW/m2 conductive geotherm n.a. not analysed

CpxCpx Al-Opx Grt-Opx Al-Opx Al-Ol PNT00 at TNT00 PBKN at TH84 PBKN at TDH10 Ni in garnet thermometry Cpx Grt-CpxGAl-Opx rt-Ol Cpx Al-Ol PNT00 at TKr88 PBKN at TOW79 PNT00 at TDH10 70 depth Estimates of equilibrium temperatures for garnets can be diamond # (km) G23 obtained from their Ni contents assuming equilibrium with G50 200 mantle olivine (Canil 1999a; Grifn et al. 1989; Ryan et al. G126 1996), which is the case for all garnet inclusions presented G201 here. Depending on the calibration used, temperatures 60 G209 range from 1175 to 1480 °C (Ryan et al. 1996 calibration)

ond 175 or 1115–1335 °C (Canil 1999a calibration). These ranges diam e graphit are smaller than the range obtained by grt–olivine Mg–Fe

P (kbar) exchange thermometry, but extend to higher T for the Ryan 50 et al. (1996) calibration. Temperature estimates for multi- 2 150 ple garnets within a single diamond generally agree within m at 38 mW/m 25 °C, compared to an uncertainty of the thermometers

adiab

2 of ± 50 °C (Ryan et al. 1996; Canil 1999a).

ntle

Kalahari geother ma The discrepancy between the two Ni–grt calibrations 40 mW/m 40 125 has been reported in the literature before, but has not been 1000 1100 1200 1300 1400 1500 resolved (Canil 1996; Grifn and Ryan 1996). One possi- bility is that the diference is due to the Ryan et al. (1996) T (°C) calibration not taking variation of Ni contents of co-existing olivine into account. As Ni contents of Akwatia olivines are Fig. 4 P T Estimated – equilibration conditions of silicate inclusions in higher than average mantle (Stachel and Harris 1997a), this Akwatia diamonds colour-coded by diamond specimen whereas dif- ferent symbols represent diferent geothermobarometer combinations may have introduced a bias towards higher temperatures. (see Table 4 for data and abbreviations, note that only diamonds with However, even when taking the diferences in Ni contents olivine inclusions have been plotted). Error ellipses represent 95% into account (3140 and 2900 µg/g Ni, respectively), the two Ni-grt confdence intervals of analytical uncertainties propagated through calibrations still difer by ca 75 °C. As T has negligible geothermobarometrical equations. Also plotted are the Kalahari geo- Can99 therm and convective mantle adiabat of Rudnick and Nyblade (1999), pressure dependence (Canil 1999a), uncertainty in equilib- 38 and 40 mW/m2 conductive geotherms (Hasterok and Chapman rium pressure cannot explain the diference. Because Ryan 2011), and the diamond–graphite transition (Kennedy and Kennedy et al. (1996) estimates resulted in temperatures for several 1976) garnets that were well above the mantle adiabat and thus unreasonably high, and the Canil (1999a) estimates agree by Wu and Zhao (2007) resulted in similar temperatures best with major-element-based P–T estimates (see previous within 50 °C. section), we used the latter calibration in the remainder of In conclusion, where P and/or T estimates could be this study. derived from multiple inclusions in a single diamond, these estimates generally agree within uncertainty of the Trace element thermometry of olivine geothermobarometer calibrations. A few inclusions are clearly anomalous, as they result in anomalous P–T esti- Aluminium, Cr and Ca in olivine have been calibrated as mates (e.g., olivine G126–156, and all three opx inclusions single-mineral geothermobarometers for garnet-bearing in diamond G111), which indicates these were not in equi- peridotites (Bussweiler et al. 2017; De Hoog et al. 2010; librium with the other inclusions in the host diamond, so Köhler and Brey 1990). However, the calibrations only were trapped in an evolving chemical environment and/or involved lherzolitic assemblages and were not tested on at diferent P–T conditions. cpx-free rocks with subcalcic garnets. This could poten- tially lead to inaccuracies, particularly for Ca, as the activ- ity of this element is no longer bufered by a Ca-saturated co-existing phase, which is essential for the accuracy of single-mineral geothermometers (De Hoog et al. 2010). In

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 15 of 28 100

150014001300 1200 1100 1000 900°C 30 10 A B

10 1.0 × 1000 × 1000 ol/gt Ca ol/gt Al D D 3 0.10

harzburgitic harzburgitic lherzolitic lherzolitic xenoliths xenoliths 1 0.01 0.10.3 1 3 0.50 0.60 0.70 0.80 0.90 ol/gt Ni-grt DAl × 1000 1000/TCanil (K)

Dol/grt Dol/grt Dol/grt Tcpx-opx Fig. 5 available, so BKN was used instead. Symbols connected by stippled Diagram ofNi-grt a Ca vs. Al ; b Al vs. equilibration tem- perature [1000/T (K)] for Al. Plus symbols represent mantle lines indicate multiple analyses of the same inclusion Canil TNi-grt xenolith data from De Hoog et al. (2010), for which no Canil was

addition, ­Cr2O3 contents of Akwatia garnets are on average 1160 °C. If the thermometer is corrected for CaO content much higher, and thus ­Al2O3 contents lower, than those of garnet, the range is much reduced (1115–1350 °C) and used for calibration of the Al–ol thermometer (De Hoog the average is considerably higher (1255 °C). Thus, Ca–ol et al. 2010). Even though the Al–ol thermometer has a temperatures without correction are signifcantly biased correction for Cr#, its validity has not been tested using towards low temperatures, but after correction they are TAl-ol TAl-ol TCa-ol the high Cr# observed in this study (0.47–0.90 compared higher than DH10 . As DH10 and DH10 have diferent P sen- to 0.33–0.75 in the mantle xenolith database). Therefore, sitivities, this diference could be due to the arbitrary P taken TAl-ol TCa-ol we compared DH10 and DH10 (De Hoog et al. 2010) using for this comparison. However, the diference is also visible (1) measured Ca and Al contents, with (2) using Ca and Al when P can be calculated, as in two harzburgitic diamonds corrected for the deviation of CaO and Al­ 2O3 of co-exist- (G209 and G126) in which olivine co-exists with garnet and ing garnets from average CaO and Al­ 2O3 in garnet from opx. In diamond G209, Al and Ca in olivine lead to P–T the calibration of De Hoog et al. (2010) (e.g., ­Cacorr = Caol/ estimates of 1170 °C at 57 kbar and 1520 °C at 93 kbar, CaOgrt × CaOavg, where CaO­ grt = CaO (wt%) in co-existing respectively. In diamond G126, in which two pairs each of garnet and ­CaOavg = average CaO (wt%) of garnet used garnets and olivines were present, there is a considerable TCa-ol TAl-ol PAl-opx in DH10 calibration, which was 5.02 wt%). Note that the range in DH10 at BKN (1205–1320 °C and 55–65 kbar) and TCa-ol PAl-opx ­Al2O3 content of garnet is inversely correlated with its an equally large range in DH10 at BKN (1475–1620 °C and TAl-ol Cr­ 2O3 content. As the DH10 formulation also incorporates 79–94 kbar). This suggests not only that not all inclusions a Cr# correction, we used the Al–ol formulation without in the latter diamond were in equilibrium, as was already Cr# correction from Eq. 13 in De Hoog et al. (2010) to shown in the section on traditional geothermometry, but also TCa- ol avoid incorporating the efect of Cr# twice. that DH10 estimates for both diamonds are unreasonably TAl-ol When using a fxed pressure of 55 kbar, the range of DH10 high. is 1005–1295 °C with an average of 1195 °C. After correct- Anomalously high Ca contents in olivine are also indi- Dol/gt Dol/gt ing for garnet ­Al2O3 content, the range is a little smaller cated by a plot of Al vs. Ca (Fig. 5a). Both these param- (1045–1280 °C) and the average is slightly higher (1205 °C), eters are similarly sensitive to temperature, so one would but well within the uncertainty of the thermometer, which expect these to correlate strongly. This is indeed the case is ± 20 °C (De Hoog et al. 2010). This indicates that the for lherzolitic mantle xenoliths (De Hoog et al. 2010), but TAl-ol efect of high Cr contents of garnet on DH10 is negligible. many Akwatia olivine inclusions fall towards the high Ca/ TCa-ol TCa-ol For DH10 the range is 980–1370 °C with an average of Al side, and will, therefore, show too high DH10 values.

1 3 100 Page 16 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

This latter may be due to the CaO correction for subcalcic of equilibrium appears justifed, but geothermometers are garnets being inaccurate (i.e., equilibration temperature is highly sensitive to small variations in concentrations of the Dol/gt not linearly correlated with Ca ), but we note that several elements involved, so even when minerals appear to be close lherzolitic inclusions (which avoid the calibration issues to equilibrium, they may still yield erroneous temperatures. mentioned above) also plot above the mantle lherzolites Also, due to diferences in their difusivities, some elements TCa-ol array. In addition, comparison of DH10 with the single cpx may be in equilibrium, whereas others are not. Tcpx thermobarometer ( NT00 ; Nimis and Taylor 2000) also gives A further thermometer based on equilibrium between TNi-grt inconsistent results. garnet and olivine is T­ Al-ol. Figure 5b shows Can99 of the Dol/gt Therefore, we suspect the high Ca/Al olivines were inclusions vs. Al . Overall, the olivine inclusions plot near afected by transient heating events at the time of their the mantle xenolith trend (based on conventional geothermo- entrapment in the diamond. Calcium difuses faster than Al barometry of four-phase lherzolites; De Hoog et al. 2010), Dol/gt TNi-grt and during heating, Ca will difuse into the olivine, resulting but the correlation between Al and Can99 is poor. Some in Ca zoning with increasing concentrations towards the oli- of the scatter is derived from heterogeneous distribution of vine rim (Köhler and Brey 1990). When these zoned olivines Al in several olivines (Fig. 3), but this does not explain the Dol/gt TNi-grt Dol/gt are trapped in diamond, they would homogenise due to dif- full extent of the range in Al at constant Can99 . Ca vs. fusion during storage in the mantle, but cannot re-equilibrate TNi-grt Dol/gt Can99 (not plotted) shows a similar scatter as Al . with other silicate minerals. This would result in high Ca/Al TCa-ol In summary, even though most geothermobarometers give ratios and anomalous DH10 values. TAl-ol similar pressure–temperature ranges on average, correlation In summary, the Cr# correction of DH10 is valid for the between diferent calibrations is poor. This is partially due high Cr# of garnet and olivine in the Akwatia suite, and to either inaccuracies based on calibrations potentially not this single-element thermometer can be used reliably for being applicable to harzburgitic paragenesis inclusions (e.g., diamond inclusion suites. However, Ca-based geothermo- TCa-ol DH10 ) or large uncertainties associated with some of the barometers (De Hoog et al. 2010; Köhler and Brey 1988) Tgrt-ol calibrations (e.g., OW79 ). However, discrepancies exceed should only be applied to olivine inclusions in diamonds that expectations even taking these factors into account. In are undoubtedly of lherzolitic paragenesis (see “Discussion” some instances, the main factor appears to be disequilibrium for criteria), but, as shown above, even for those samples, between mineral inclusions in the same diamond, which sug- results are often still unreliable. gest that the diamonds grew in an environment with evolv- ing P–T conditions and/or chemistry (e.g., Bulanova 1995). From this perspective, single-mineral calibrations (cpx; Comparison between major and trace‑element Pcpx Tcpx NT00 at NT00 ) or calibrations that are only weakly reliant based geothermometers TNi-grt TAl-ol on co-existing mineral composition (e.g., Can99 , DH10 ) are therefore preferable when studying diamond inclusion suites. Nickel in garnet temperatures show very little correlation with Mg–Fe exchange garnet–olivine temperatures, despite Thermal structure of diamond source mantle both thermometers being based on the same mineral pairs. Only about half the values fall within 50 °C of each other. Tgrt-ol TNi-grt Taking into account the uncertainties in geothermometry The OW79 range is much larger than the Can99 range (460 vs. results discussed earlier, the Akwatia dataset can be used Tgrt-ol 220 °C, respectively) and several values of OW79 are outside to provide constraints on the thermal structure of the man- the diamond stability feld (< 1050 °C) or above the mantle tle beneath the West Africa Craton at the time of diamond adiabat (> 1360 °C), indicating that these T estimates are formation (> 2.2 Ga). In total, fve diamonds were available unreliable. Only the average of both thermometers agrees for which pressure estimates could be calculated (opx or cpx Tgrt-ol well (1215 °C and 1210 °C, respectively), but for OW79 this present), and six diferent thermometer–barometer combina- value is dependent on the pressure assumed in the calibration tions were used to calculate P–T estimates (Fig. 4). When (55 kbar in this study). multiple inclusions of the same type were present, separate There are several possible explanations for the discrepan- P–T estimates were calculated for each inclusion. Consid- Tgrt-ol TNi-grt cies between OW79 and Can99 . Both methods assume equi- erable scatter is observed, with variation in T estimates for librium between garnet and olivine inclusions from the same single diamonds ranging from 100 up to 300 °C, whereas diamond, whereas in reality they may have become encap- the minimum pressure variation is 4 kbar, but can be as sulated during diferent stages in the growth of their host high as 16 kbar. Some of the variation may be explained diamond that were associated with fuctuations in the chemi- by uncertainty in the P–T calculations, which can be large TAl-ol cal environment or formation temperature. As Ti contents especially for pressure-sensitive thermometers (e.g., DH10 ) PAl-opx of garnet and olivine correlate well (Fig. 7), the assumption and temperature-sensitive barometers (e.g., BKN ). The

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 17 of 28 100 propagated analytical uncertainties (not considering errors inclusions fall within the diamond stability feld and below the inherent in the calibration of the used geothermobarometers) mantle adiabat. A 39 mW/m2 also results in an acceptable P–T are indicated in Fig. 4 by error ellipses, and even with our range, whereas using cooler conductive geotherms results in high-precision EPMA data, this can lead to uncertainty in temperature estimates above the mantle adiabat. Thus, based TAl-ol pressure estimates of up to 10 kbar. Due to its low tempera- on DH10 data alone, we can determine the paleogeotherm to be Pcpx 2 ture sensitivity, NT00 gives much smaller uncertainties on a 38–39 mW/m conductive geotherm. This is the same result individual P–T calculations than other geobarometers, but as obtained by traditional geothermobarometry. like other geothermometers gives inconsistent results; the Pcpx P–T range for each diamond using NT00 is no smaller than when using PBK90 estimates. Discussion Nevertheless, several observations can be made. First, as one would expect, all P–T estimates fall within the diamond Diamond paragenesis stability feld. Second, the paleogeotherm of the West Africa Craton is somewhat cooler than the Kalahari geotherm (based Peridotitic inclusions in diamonds can be subdivided into on the Kaapvaal and Zimbabwe cratons; Rudnick and Nyblade lherzolitic and harzburgitic parageneses (representing man- 1999) and scatters around 38–40 mW/m2 conductive geo- tle source lithologies) based on the presence or absence of therms (Hasterok and Chapman 2011), in agreement with the cpx, or Ca-saturation of garnet (Sobolev et al. 1973; Stachel smaller dataset from Stachel and Harris (1997b). This scatter and Harris 2008). Harzburgitic assemblages with strongly could refect actual temporal or spatial variations in heat fow Ca-depleted garnets (G10 garnets; Dawson and Stephens beneath the West Africa Craton, but note that several diamonds 1975; Grütter et al. 2004) are of particular interest, as these (G50, G201) cover nearly the whole range of heat fow esti- are indicators for diamondiferous kimberlites (Gurney 1984; mates. Third, the diamonds of the Akwatia feld were derived Gurney et al. 1993). However, many diamonds contain oli- from a continuous mantle section from the diamond–graphite vine inclusions only and no criteria exist to determine their boundary (ca. 140-km depth) down to ca. 200 km. This latter paragenesis. Here we evaluate diamond paragenesis based depth is the expected thickness of the lithosphere for a heat on the trace element composition of its olivine inclusions. fow of 38–39 mW/m2 (Hasterok and Chapman 2011). Similar estimates can be derived from olivine data alone, Ca–Al systematics TAl- ol as DH10 is insensitive to mantle lithology and gives simi- TAl-ol lar results to other geothermometers. However, DH10 alone A strong positive correlation exists between Ca and Al in Ca-ol does not give an accurate P–T estimate. PKB90 has been cali- grt–lherzolite mantle olivine due to the temperature depend- brated as a geobarometer (Köhler and Brey 1990; De Hoog ence of concentrations of both elements in olivine and rela- TAl-ol et al. 2010), but combining this with DH10 gives unreason- tively constant CaO and ­Al2O3 activities in co-existing ably large uncertainties due to the shallow intercept of the garnet and cpx (De Hoog et al. 2010). Indeed, lherzolitic two geothermobarometers in P–T space. As an alternative, for olivines from Akwatia plot close to the mantle xenolith sufciently large suites of diamond-hosted olivines the range Ca–Al array (De Hoog et al. 2010, Sobolev et al. 2009), TAl-ol of DH10 values can be used to determine the thermal state with two exception of two inclusions, which plot towards of the mantle at the time of olivine entrapment. Essentially, the high Ca/Al side. TAl-ol by calculating DH10 using pressure from a certain conductive In contrast to lherzolitic olivines, Ca and Al contents geotherm T, a range of P–T conditions is calculated which of harzburgitic and peridotitic (paragenesis undetermined) can then be compared with the stability feld of diamond Akwatia olivines show signifcant scatter (Fig. 6a), most and the mantle adiabat temperature as additional constraints. having low Ca/Al values. This scatter could be caused by TAl-ol Calculating DH10 for all olivines from the Akwatia inclusion olivines equilibrating along diferent geothermal gradients suite using a 40 mW/m2 conductive geotherm of Hasterok and the diference in pressure dependence of Ca and Al and Chapman (2011) results in a T range of 1005–1295 °C partitioning (De Hoog et al. 2010; Köhler and Brey 1990). (ignoring one outlier of 930 °C), whereas the 40 mW/m2 geo- However, although the exact origin of the Akwatia suite of therm intercepts the diamond–graphite transition intercepts at alluvial diamonds is unclear, they are likely of similar age TAl-ol 1100 °C and 46 kbar. Thus, the DH10 data is incompatible and unlikely to refect large temporal variations in geother- with a 40 mW/m2 conductive geotherm, as it would imply mal gradients needed to signifcantly shift Ca–Al trends that some olivine inclusions were trapped inside the graph- (see Ca–Al correlation for various conductive geotherms; ite stability feld. Using a 38 mW/m2 conductive results in a Fig. 6a). More likely, Ca contents in olivine refect low but range of 1042–1355 °C (ignoring one outlier), whilst it inter- variable Ca activity of the mantle environment, as indicated cepts the diamond–graphite transition at 950 °C and 44 kbar by CaO undersaturation of co-existing garnets (Stachel and and the mantle adiabat at 1370 °C. Thus, P–T estimates of all Harris 1997b). A large fraction of low-Ca garnets was also

1 3 100 Page 18 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

700 350 A B 600 300 Lhz 500 Kalahari geotherm 250 ) ) pm pm 400 40 mW/m2 200

300 Na (p 150 Ca (p Per 38 mW/m2 Hrz Lhz 200 100 36 mW/m2 harzburgitic harzburgitic lherzolitic lherzolitic 100 50 peridotitic Hrz peridotitic xenoliths xenoliths 0 0 0150 00 1502200350 00 0150 00 1502200350 00 Al (ppm) Al (ppm)

1200 400 C 9 harzburgitic D 350 lherzolitic 1000 7 peridotitic Cr# = 0. xenoliths 300 Cr# = 0. 800 )

) 250

600 200 ( ppm ( ppm

Cr .5 Na 150 400 Cr# = 0 Lhz 100 harzburgitic 200 lherzolitic 50 peridotitic xenoliths Hrz 0 0 0150 00 200 250 300 0100 200 300 400 500 600 700 Al (ppm) Ca (ppm)

Fig. 6 Ca vs. Al (a), Na vs. Al (b), Cr vs. Al (c) and Ca vs. Na (d) geotherms, including the Kalahari geotherm of Rudnick and Nyblade concentrations in olivine inclusions in Akwatia diamonds. Sym- (1999) and 36, 38 and 40 mW/m2 geotherms from Hasterok and bols connected with stippled lines indicate multiple analyses of the Chapman (2011) calculated by inverting TAl-ol and TCa-ol equations same olivine inclusion. Compositions of olivines from the literature from De Hoog et al. (2010) with Cr#ol = 0.6, with 50 °C temperature are plotted for comparison (black plus and cross signs: grt lherzolite intervals (open circles) and the highest temperature indicated being and harzburgite xenoliths with equilibration temperatures > 1050 °C, 1350 °C, the lowest temperature being where each geotherm inter- respectively, De Hoog et al. 2010; small red diamonds and small blue sects the graphite–diamond boundary. Fields indicated with dashed plus signs: grt peridotite xenoliths and diamond-hosted inclusions, lines in (a), (b) and (d) indicate paragenesis of olivine inclusions in respectively, Sobolev et al. 2009). Further indicated in (a) is the pre- diamonds, with Lhz = lherzolitic paragenesis, Hrz = harzburgitic and dicted Ca–Al trend in olivines equilibrated along various conductive Per = peridotitic (undetermined paragenesis) observed in the diamond inclusions dataset from Sobolev with Ca/Al < 2.2 are of harzburgitic origin (Fig. 6a). In addi- et al. (2009). tion, since olivine in equilibrium with clinopyroxene has at The different location of lherzolitic and harzburgitic least 100 µg/g Ca in the P–T range where diamond is stable olivines in Al–Ca space allows olivine compositions to be (> 1050 °C; Fig. 6a), olivines with < 100 µg/g Ca must also used to determine host diamond paragenesis. All olivines be harzburgitic, even if they fall on the mantle olivine Al–Ca

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 19 of 28 100 array. Note that this applies to olivine inclusions in diamonds Al > 0.75 are lherzolitic, whereas olivines with Na ≤ 60 µg/g only, as low-T lherzolite xenoliths also have Ca < 100 µg/g, and Na/Al ≤ 0.75 are harzburgitic (Fig. 6b). Following these but these were sourced from the graphite stability feld (De rules, all 19 harzburgitic inclusions are correctly classifed Hoog et al. 2010). All olivines with > 300 µg/g Ca and Ca/ as such, and so are all fve lherzolitic inclusions. Of the four Al ≥ 2.2 are of lherzolitic paragenesis. For olivines that fall peridotitic olivine inclusions (paragenesis undetermined), on or near the mantle array and have 100–300 µg/g Ca, their three are classifed as harzburgitic, whereas the remaining paragenesis is uncertain (i.e., peridotitic). Following these one is uncertain due to heterogeneity in its Al and Na con- rules, 12 out of 19 harzburgitic olivine inclusions are cor- tents. Na–Ca systematics show similar patterns, so olivine rectly classifed as such, whereas seven cannot be classifed. inclusions with < 60 µg/g Na and < 300 µg/g Ca are harz- Of the four peridotitic olivine inclusions (paragenesis unde- burgitic (Fig. 6d). termined), two are harzburgitic according this classifcation. Combining the Ca–Al, Na–Ca and Na–Al classifcations, All fve lherzolitic inclusions are identifed correctly. all harzburgitic and lherzolitic inclusions are classifed suc- As the distinction is based on generally applicable trace- cessfully, and all peridotitic (unspecifed) inclusions were element partitioning data, these rules can also be applied to classifed as harzburgitic. Harzburgitic olivines dominate the other diamond-hosted inclusions suites. When applied to the Akwatia suite, which matches the distribution of the Akwa- diamond-hosted inclusion suites from Sobolev et al. (2009), tia garnet inclusions, as well as the distribution of diamond- the majority of olivines from Yakutia mines (299 out of 505, hosted inclusions suites worldwide (Gurney 1984; Stachel or 59%) are harzburgitic, whereas only 20 are lherzolitic and Harris 2008). (4%), the remainder being of unidentifed origin (perido- titic). In contrast, olivines from the Majhgawan mine, India, Implications for olivine as a diamond indicator mineral and from Argyle, Australia, as well as the majority of inclu- sions from Snap Lake, Canada, cannot be classifed (i.e., are Olivine can potentially be used as a diamond indicator min- peridotitic). As no Na data were reported, further refnement eral. Previous use of olivine in diamond exploration has using the scheme below based on Na-Ca and Na-Al sys- mainly focussed on identifying olivine from kimberlite, i.e., tematics is unfortunately not possible for the Sobolev et al. as a kimberlite indicator mineral (e.g., Matveev and Stachel (2009) dataset. 2007; Shchukina and Shchukin 2018), but this provides little indication of diamond potential. Diamond formation is asso- Na–Al–Ca systematics ciated with garnet-bearing sections of the SCLM in which harzburgites predominate over lherzolites (Gurney et al. Although many olivine inclusions can be successfully clas- 1993). Olivines with a harzburgitic paragenesis, following sifed using Ca–Al systematics, a signifcant proportion of the identifcation scheme presented earlier, are derived from harzburgitic inclusions could not be classifed due to overlap the same source as G10 garnets. However, since we only of harzburgitic and lherzolitic inclusions (peridotite feld in compared the olivine inclusions to olivines from garnet peri- Fig. 6a). As harzburgites are depleted in Na as well as Ca, dotite xenoliths with equilibrium temperatures > 1050 °C, it Na–Al–Ca systematics may be used to provide additional is possible that olivines from lower temperature xenoliths, constraints on mantle paragenesis. Sodium shows well- or from other lithologies, have Al–Ca–Na systematics that defned trends with Al and Ca in olivines from lherzolitic would identify them as harzburgitic, whereas in fact they mantle xenoliths, as Na activity is bufered by co-existing could be derived from, e.g., low-T lherzolites. Indeed, we clinopyroxene and therefore, like for Ca and Al, tempera- note that a few lherzolitic mantle xenoliths plot inside the ture is the main control on Na contents of olivine (De Hoog felds we have identifed as harzburgitic (Fig. 6). These et al. 2010). In contrast, Akwatia olivines have very low Na are xenoliths that have equilibrium temperatures just over contents relative to Al (Fig. 6b) and Ca (Fig. 6d). Even most 1050 °C, and thus are very close to the graphite–diamond lherzolitic olivine inclusions fall below the xenolith-based transition. In addition, olivines from low-T lherzolite xeno- Al–Na and Ca–Na arrays of grt–lherzolites (De Hoog et al. liths would overlap with our harzburgite feld as well (De 2010). Akwatia clinopyroxene inclusions have low jadeite Hoog et al. 2010). Thus, more stringent fltering needs to contents (1–3%) compared to clinopyroxene from mantle be applied to be certain that an olivine of unknown origin is xenoliths (4–14%; De Hoog et al. 2010) and clinopyroxene derived from the same source as G10 garnets. in diamonds elsewhere (7 ± 4%; Stachel and Harris 2008), We recommend that the dataset is frst fltered to olivines which probably causes the relatively low Na contents of lher- of mantle origin (Fo = 92–95% and NiO > 0.3 wt%; Fig. 2) zolitic olivine inclusions from Akwatia. Harzburgitic olivine with Al > 40 µg/g, Ca < 300 µg/g and Na < 60 µg/g. This inclusions have even lower Na contents. Therefore, Na–Al removes most lherzolites and low-T peridotites, as well systematics allow us to distinguish between harzburgitic as olivines of non-mantle origin. Then ‘G10’ olivines are and lherzolitic olivines: olivines with Na > 60 µg/g or Na/ those from the fltered dataset that have at least two of the

1 3 100 Page 20 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

olivine inclusions in diamond mantle xenoliths harzburgitic lherzolitic peridotitic grt-lhz sp-lhz SSZ sp-hrz metasomatised 1000 A B

100 )

10 ivine (ppm

ol refertilization in

Ti 1

melt depletion

0.1 110100 1000 0.00.2 0.40.6 0.81.0 Ti in garnet (ppm) Cr# (olivine)

Fig. 7 Co-variation of Ti in olivine vs. Ti in garnet (a) and Ti in oli- supra-subduction zone (SSZ) spinel harzburgites from Kamchatka vine vs. Cr# of olivine (b) for inclusions in Akwatia diamonds. Com- (Ionov 2010) and metasomatised mantle xenoliths (, wehrlites positions of olivine in grt and grt-spinel peridotite xenoliths from and websterites) from the Kimberley kimberlite (Rehfeldt et al. 2008) De Hoog et al. (2010) are indicated for comparison, as well as (in b) following three ratios: Ca/Al < 2.2, Al/Na > 1.5 and Ca/ diamond indicator mineral needs further evaluation, but Na < 5. This results in a proportion of harzburgitic inclusions including Na in analytical setups would be benefcial as it not being identifed as such, but no olivines from mantle is more useful to constrain diamond paragenesis than Ca. xenoliths are incorrectly identifed as harzburgitic. Apply- ing this scheme to the 28 Akwatia olivine inclusions results Formation conditions of diamond mantle source in 11 ‘G10’ olivines, whereas none of the olivines from De Hoog et al. (2010), including garnet and spinel peridotites, Melt depletion and refertilisation kimberlites and picrites, were classifed as ‘G10’ olivines. If Na has not been measured, as is the case for much of The Ti content of olivine serves as a proxy for bulk rock the literature, including the high-precision EPMA tech- ­TiO2 content (De Hoog et al. 2010; Hermann et al. 2005), nique employed by several groups (Batanova et al. 2015; which is an important indicator of melt depletion and refer- Sobolev et al. 2008), the fltering can be simplifed to tilisation of mantle rocks (e.g., Bizimis et al. 2000, Rehfeldt Al > 40 µg/g, Ca < 300 µg/g and Ca/Al < 2.2. This results et al. 2008). Low Ti contents of Akwatia olivines, particu- in less olivines identifed positively as ‘G10’ olivines. As larly the harzburgitic inclusions, indicate that their mantle an example, when applied to the diamond-hosted inclu- source was highly depleted in fusible elements. Titanium sion suites from Sobolev et al. (2009), for Yakutian kim- contents in olivine correlate well with TiO­ 2 contents of co- berlites just over one-third of olivine inclusions is classi- existing garnet (Fig. 7a), although they are ofset from the fed as ‘G10’. This proportion drops considerably for the mantle xenolith array (discussed below). The low Ti contents other localities (diamonds from elsewhere in Russia, as of the garnets were taken as evidence by Stachel and Har- well as Canada), and to none at all for diamond-hosted ris (1997a) that the Akwatia diamond source rocks avoided olivine inclusions from India and Australia. Unfortunately, refertilisation through a melt. The olivine data support this due to Na missing from the Sobolev et al. (2009) dataset, view. About half the harzburgitic olivines have < 1 µg/g it is impossible to determine if the latter suites are lher- Ti, which equates to < 4 µg/g in the bulk rock (De Hoog zolitic or harzburgitic. Of all peridotite xenoliths from the et al. 2010), and the lowest values are 0.2 µg/g Ti. These Ti same Yakutia, only one contains ‘G10’ olivine. Therefore, contents are lower even than those in olivine from highly the usefulness of this olivine classifcation scheme as a depleted supra-subduction zone peridotites (0.8–2 µg/g;

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 21 of 28 100

Ionov 2010; Tollan et al. 2017) and indicate extremely high in which over a quarter of inclusions has Cr# < 0.45. This degrees of melting in excess of 30%. Similarly low Ti con- is again consistent with the mantle source of Akwatia dia- tents were reported for olivine in grt-spinel lherzolites from monds having high Cr#, and a very high degree of melt Kaalvallei, Kaapvaal Craton (De Hoog et al. 2010). In con- depletion. Importantly, lherzolitic inclusions have similar trast, Ti contents of lherzolitic olivine inclusions are similar Cr# as harzburgitic inclusions (Fig. 7b), which suggests to those of olivine from grt–lherzolite xenoliths, indicating these also experienced large degrees of melt depletion, that either their source rocks did not experience extensive which is at odds with their high ­TiO2 contents. This indi- melt depletion or that they were refertilised by Ti-bearing cates that lherzolitic olivine was afected by small degrees melts before entrapment of the inclusions into their host of metasomatism by Ti-bearing melts prior to entrapment diamonds. in the diamond, which would increase Ti contents of the One peculiar feature of the Akwatia dataset is that Ti strongly depleted mantle, including olivine, but would not in olivine vs. garnet appears to be ofset relative to the afect Cr# signifcantly (De Hoog et al. 2010; Rehfeldt mantle xenolith data (Fig. 7A), i.e., Ti in olivine is about et al. 2008). Harzburgitic Akwatia olivine inclusions four times lower than expected based on Ti in co-existing escaped refertilisation (as indicated by their low Ti and garnet. This is true for lherzolitic as well as harzburgitic high Cr#, Fig. 7b), presumably due to isolation from the olivine. This ofset is difcult to explain as one would surrounding mantle by the diamond host, and consequently Dgt/ol expect Ti to be relatively constant, with only a small were encapsulated into their host diamonds prior to any dependence on temperature and garnet composition. A major refertilisation. change in Ti substitution from the octahedral to the tetra- A similar pattern of ­TiO2 vs. Cr# can be seen for Akwatia hedral site was reported in wet versus dry mantle olivine, garnet inclusions, but Cr# tend to be lower for lherzolitic and the presence of Na appears to enhance uptake of Ti garnets than harzburgitic ones, and would therefore also be in olivine (Tollan et al. 2018). As Na contents in Akwatia consistent with a partial melting trend for lherzolitic garnets. olivines inclusions are relatively low, we cannot exclude This diference in behaviour between olivine and garnet is a crystal-chemical control on Ti partitioning. Neverthe- likely due to the poor correlation between Cr# of olivine and less, the broad correlation with of Ti in Akwatia olivine garnet inclusions from the same diamond. This is a common inclusion with TiO­ 2 in co-existing garnet does indicate that feature also noted in mantle xenoliths (De Hoog et al. 2010), Ti in olivine does follow a similar pattern as the mantle where a much better correlation between the Cr# of olivine xenolith database and refects that of the bulk rock (De and cpx was observed than with garnet. This is probably Hoog et al. 2010). due to the pressure dependence of Cr# in garnet in equilib- To investigate whether lherzolitic olivines were referti- rium with Cr-spinel (Grütter et al. 2006) and the fact that the lised by melt infltration or instead escaped melt depletion Cr# of spinel in equilibrium with garnet increases further altogether, we compared olivine TiO­ 2 contents with their with depth until spinel becomes exhausted (Klemme 2004). Cr# (Fig. 7b). Cr# of olivine serves as an indicator of melt Confrmation that lherzolitic inclusions probably record depletion is, as it correlates with Cr# of co-existing chro- melt refertilisation comes from cpx inclusions in the Akwa- mite (De Hoog et al. 2010), which increases with increas- tia diamonds. These have Cr# from 27 up to 41%, which is ing degrees of melt extraction (e.g., Dick and Bullen 1984; considerably higher than cpx from strongly depleted supra- Hellebrand et al. 2001). This is also illustrated by compar- subduction zone harzburgitic xenoliths (Ionov 2010), which ing Ti and Cr# of fertile spinel lherzolites with residual have Cr# ranging from 16 to 25%. Therefore, the high Cr# of supra-subduction zone harzburgites (Fig. 7b). It is thought Akwatia cpx is due to their formation during melt infltration that most garnet peridotites from the subcontinental litho- with the Cr# of cpx being imposed by the composition of sphere underwent low-pressure melting in the spinel sta- the ambient mantle. bility feld (Brey and Shu 2018; Stachel et al. 1998) and would, therefore, show a similar Cr#–TiO2 trend. Cooling Trivalent cation behaviour and mantle oxidation state after melting would not change Cr# of olivine much, as Al and Cr have similar temperature and pressure dependence Vanadium and Cr can occur in several oxidation states and (De Hoog et al. 2010). Most Akwatia olivines, includ- may, therefore, provide information on the redox condition ing lherzolitic inclusions, have Cr# of 0.6–0.7 with some of their environment (e.g., Papike et al. 2005). Vanadium higher Cr# values up to 0.9 (Fig. 6c). This suggests that and Cr in Akwatia olivine inclusions (Fig. 8a) are rather the mantle source of the olivines had very high Cr#, con- scattered instead of the excellent correlation observed in sistent with the high Cr­ 2O3 contents of garnet inclusions in olivines from mantle xenoliths (De Hoog et al. 2010; Ionov the Akwatia diamonds (Stachel and Harris 1997a). No Cr# 2010; Witt-Eickschen and O’Neill 2005), showing high and values lower than 0.45 are present, in contrast to the dia- variable Cr/V values instead. Akwatia harzburgitic garnets mond-hosted inclusion suites from Sobolev et al. (2009), also exhibit high Cr/V compared to garnets from mantle

1 3 100 Page 22 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

olivine inclusions in diamond mantle xenoliths

lherzolitic peridotitic harzburgitic low-T GP lhz high-T GP lhz SP lhz SSZ SP hrz

0.05 A B 1000

0.01 t r l/gr o 100 C D Cr (µg/g)

0.001 harzburgitic lherzolitic xenoliths 10 1.05.0 10.0 800100012001400

V (µg/g) T°C (Ni-Grt; BKN)

300 30 C D

25

100 20

50 15 (atoms pfu) r] Al (µg/g)

C 1:1 + l 10 A #[

10 5

5 0 1.010.0 0510 15 V (µg/g) #Na (atoms pfu)

Dol/grt TNi-grt Fig. 8 Plots of a Cr vs. V, b Cr vs. Canil (°C), c Al vs. V, and d 2010), spinel lherzolites (De Hoog et al. 2010; Witt-Eickschen and O’Neill 2005) and supra-subduction zone spinel harzburgites (Ionov #Na vs. #[Al + Cr] in olivines from Akwatia diamonds compared to TNi-grt Tcpx-opx olivine from mantle xenoliths including garnet peridotites, separated 2010). For the mantle xenoliths (b) no Canil was available, so BKN in high- and low-T peridotites (boundary = 1050 °C) (De Hoog et al. was used instead

xenoliths, in which both elements have a narrow composi- are controlled by fO2. Under typical oxidation conditions tional range in their major hosts cpx and garnet, and their in the upper mantle (IW to NNO bufers; Berry et al. 2003) concentrations in olivine are largely controlled by equili- most V will occur as V­ 3+ or V­ 4+ and Cr as Cr­ 3+ (Canil bration temperature of the xenoliths (De Hoog et al. 2010). 2002; Papike et al. 2005). V­ 4+ has an ionic radius close Based on their equilibration temperatures, Akwatia olivine to ­Ti4+, which has very low compatibility in olivine. V­ 2+ inclusions are expected to have > 5 µg/g V and > 150 µg/g would probably be highly compatible in olivine, having an Cr, but nearly all harzburgitic olivine inclusions have lower ionic radius very close to ­Fe2+, but occurs only at very low V. Chromium in olivine is also lower than expected when oxidation states (below the IW bufer). Cr­ 2+ has a simi- considering the high Cr# of co-existing garnets. lar ionic radius and charge to ­Fe2+ and ­Mn2+, and there- One possible explanation for the offset of Akwatia fore expected to be more compatible in olivine than Cr­ 3+. inclusions from the mantle array is that Cr and V contents Indeed, in peridotite melting experiments, olivine-melt

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 23 of 28 100 partition coefcients for Cr­ 2+ and Cr­ 3+ are comparable, In addition, despite the high Cr contents of some Akwatia Dol/opx Dol/cpx but Cr and Cr increase strongly with lower oxygen olivines, most olivines have in fact relatively low Cr contents fugacity (e.g., Mallmann and O’Neill 2009). A similar considering their equilibration temperature and high Cr# in increase in compatibility of ­Cr2+ in olivine at low oxida- co-existing garnet. This is demonstrated in Fig. 8b., where Dol/grt tion fugacity was observed in Cr-spinel bearing lithologies Cr ranges from 0.001 to 0.009 for harzburgitic olivine (Li et al. 1995). inclusions in Akwatia diamonds, compared to 0.005–0.016 Based on the crystal-chemically induced relationship for olivine from the mantle xenolith database equilibrated between oxygen fugacity and pressure, in average lith- at similar temperatures. A similar pattern of low Dol/grt val- ospheric mantle fO2 decreases from about ΔQMF = − 1 to ues can be observed for V and Sc (not shown), but not for − 2.9 between 100- and 200-km depth (Canil et al. 1994), Al (Fig. 5b). As Sc is not afected by oxidation state but i.e., remains well above the IW bufer and the stability feld otherwise has a similar geochemical behaviour as Cr and V, of ­V2+ (Canil et al. 1994; Luth and Stachel 2014; Stagno any redox-sensitive processes should be clearly visible in et al. 2013). fO2 estimates based on Cr-spinel–olivine pairs changes in Cr/Sc and/or V/Sc (Canil 1999b, 2002). No such in Akwatia diamonds (Cr-spinel compositions from Stachel diference is observed for V, and only a small diference for and Harris 1997b) using the Ballhaus et al. (1991) calibra- Cr/Sc, probably related to the presence of divalent Cr, as tion (with 1994 correction; Ballhaus et al. 1994) average discussed above. to ΔQMF = − 1.6 ± 1.3. This is close to the EMOD bufer Thus, Cr, Sc and V all have low DOl/grt values for harz- (enstatite + magnesite = olivine + diamond), and probably burgitic olivines, whereas lherzolitic inclusions are similar an overestimate of oxygen fugacity due to inaccuracies or higher than olivines from mantle xenoliths (Fig. 8b). We of the Ballhaus et al. (1991) calibration at high pressures. interpret the diference between harzburgitic and lherzolitic Nevertheless, even at this relatively high oxidation state, inclusions as a change in exchange mechanisms for triva- some Cr2+ would likely be present in Akwatia olivines due lent cations due to the exhaustion of clinopyroxene. Triva- to the high temperatures of equilibration and high MgCr­ 2O4 lent cations in olivine need to be charge balanced, e.g., by contents of syngenetic chromite (Berry et al. 2003; Li et al. monovalent ions or vacancies on the M site (where Mg and 1995; Stachel and Harris 1997b). Three Akwatia diamonds Fe reside) or by trivalent ions on the Si site (De Hoog et al. with olivine as well as Cr-spinel inclusions can be used to 2010; Kurosawa et al. 1997; Witt-Eickschen and O’Neill estimate the ­Cr2+ content following Li et al. (1995). Using 2005). A strong correlation between #Na (nr. of atoms per TAl-ol average DH10 of 1140 °C and ΔQMF = − 2, ca. 80–130 µg/g functional unit) and #[Al + Cr] close to 1:1 was observed Cr­ 2+ is divalent out of 117–327 µg/g total Cr present in these for lherzolitic mantle olivine, suggesting that Na plays an olivine inclusions. These estimates suggest that a signifcant important role in the charge balance of trivalent cations in fraction of ­Cr2+ is present in most Akwatia olivines, which is garnet lherzolites (De Hoog et al. 2010). In harzburgites, in agreement with spectroscopy measurements of Cr oxida- where cpx has been exhausted during partial melting, Na tion state in olivine inclusions from South African diamonds activity in the residue has decreased strongly. Indeed, (Sutton et al. 1993), in which one-third of total Cr was pre- for Akwatia olivines, #Na is strongly depleted relative to sent as divalent Cr. This ­Cr2+ would add to total Cr in oli- #[Cr + Al], indicating that other substitution mechanisms vine, as it is incorporated following a exchange reaction dif- must be active in Na-undersaturated olivine (Fig. 8d). ferent to Cr­ 3+. Indeed, some Akwatia olivines have very high The diference in substitution mechanisms is further illus- Cr contents (up to 1090 µg/g Cr or 0.16 wt% ­Cr2O3). High trated in Fig. 8a, c and d. Chromium and V behave simi- Cr contents have also been observed in olivine inclusions larly in garnet peridotite and spinel peridotite xenoliths. from other diamond suites (Hervig et al. 1980b; Sobolev Lherzolitic olivine inclusions fall on that trend (Fig. 8a), et al. 2009). More reducing conditions are required to get or slightly above, which can be explained by the presence higher Cr contents, and must be around QMF-3 to reach 0.16 of ­Cr2+. However, all other olivine inclusions are strongly wt% observed in some Akwatia olivine inclusions, which is depleted in V compared to mantle xenoliths equilibrated still within the range of oxidation states in cratonic mantle under similar P–T conditions (high-T GP lherzolites in (Luth and Stachel 2014). We note, however, that the Li et al. Fig. 8a). Low V contents are difcult to explain with a (1995) calibration is only valid for olivine in equilibrium change in oxidation state, as ­V2+ would, like Cr­ 2+, become with Cr-spinel. Many olivines probably had no Cr-spinel in more compatible in olivine relative to co-existing their assemblage when they were trapped in the diamond, as and spinel (Li et al. 1995; Mallmann and O’Neill 2009), the 39 mW/m2 conductive geotherm intersects the Cr-spinel and requires oxygen fugacities below the IW bufer, which out boundary at about 1150 °C and 52 kbar (Klemme et al. are unlikely to occur in cratonic mantle (Luth and Stachel 2009), whereas most olivine inclusions were derived from 2014). Therefore, we conclude that the lack of Na for charge higher P–T conditions. Thus, the amount of divalent Cr in balance reduces the amount of V in harzburgitic olivine. Fig- olivine from Cr-spinel-free assemblages remains uncertain. ure 8d shows that #Na is signifcantly lower than the amount

1 3 100 Page 24 of 28 Contributions to Mineralogy and Petrology (2019) 174:100 required to charge balance trivalent cations (expressed as have slower ascent rates than kimberlites (e.g., Rus- #[Al + Cr], note that for simplicity of the discussion we have sell et al. 2012). The following reaction was proposed by ignored Li, Sc and V in this fgure, as their contribution is Ingrin and Skogby (2000) to explain possible hydrogen loss insignifcant relative to Na, Al and Cr), and therefore other from mantle minerals during ascent to the surface: substitution mechanisms than Na–M3+ must play a role. Fe2+ + OH− = Fe3+ + O2− + 1∕2H It is worth noting that of the trivalent cations, Al par- 2 (1) titioning into olivine appears to be least afected by the 2+ but if ­Cr is present, this reaction would be preceded by: absence of cpx, which is important for its use as a single- TAl-ol 2+ − 3+ 2− element thermometer ( DH10 ). Its concentrations are similar Cr + OH = Cr + O + 1∕2H2 (2) in harzburgitic and lherzolitic inclusions, as well as olivines 2+ 2+ from mantle xenoliths (Fig. 8c). It is also relatively unaf- as ­Cr oxidises at lower oxygen fugacity than Fe­ fected by the presence of spinel, which has a strong efect (Schreiber et al. 1987). As hydrogen difuses very fast in on the distribution of Cr and V (Fig. 8a, c). Whilst in garnet olivine and, therefore, readily escapes into the boundary peridotites the Na–Al exchange mechanism appears to be layer between olivine and diamond, this process results in dominant (De Hoog et al. 2010), in spinel peridotites a Tsch- oxidation of the olivine inclusion. The presence of methane ermak-style substitution of Al occurs, i.e., Al on the M site ­(CH4), detected in thin fuid layers around silicate inclusions is not charge balanced by Na, but by another Al replacing in diamonds, has been interpreted as the result of ­H2 escap- Si (De Hoog et al. 2010; Witt-Eickschen and O’Neill 2005). ing from the inclusions and reacting with the diamond host It appears that Tschermak substitution becomes important to form CH­ 4 (Smith et al. 2018). It also explains the observa- in garnet harzburgites if little Na to charge-balance Al is tion that no hydroxyl has been detected in diamond-hosted present. olivine inclusions (Matsyuk and Langer 2004), whereas the maximum solubility of H­ 2O in olivine equilibrated at Chromium heterogeneity: syn‑eruptive hydrogen 5–6 GPa is ca. 150–200 ppm (Demouchy and Bolfan-Casa- loss? nova 2016). This amount of ­H2O is enough (on a 1:1 atomic equivalent) to oxidise 870–1160 µg/g ­Cr2+, which is higher than ­Cr2+ present in the olivine inclusions, as calculated in A complicating factor to the interpretation of V and Cr con- the section on ‘Mantle oxidation state’ above. tents of Akwatia olivines is the large within-grain variability 3+ As the solubility of ­Cr in the oxidised olivine is much observed for both elements, whereas most other trace ele- 2+ lower than that of ­Cr (Li et al. 1995), the reaction may ments show no such variability (Fig. 3). Complex oscilla- lead to local oversaturation of Cr, resulting in precipita- tory zoning patterns of P, Cr and Al have been recorded in tion of chromite, which would also afect the distribution olivines from basaltic lavas and spinel peridotite xenoliths of Al and V, both of which are more compatible in chromite (Mallmann et al. 2009; Milman-Barris et al. 2008), whereas than olivine. However, no chromite micro-inclusions were other elements showed no such zoning (note that V was not observed by BSE imaging (with a maximum resolution of measured in these studies). This indicates slow difusion of ca. 100 nm), and sub-µm chromite inclusions are generally Cr and Al in olivine relative to Ca, Mn and Ni and thus extremely rare in olivine inclusions in diamonds (Sobolev points to Cr–Al zoning in Akwatia olivine inclusions being et al. 2008). The spatial distribution of Cr (Electronic Sup- a primary growth feature. However, Akwatia olivines do not plement, Fig. S4) shows that its distribution is highly hetero- show oscillatory zoning, as element mapping shows that Cr geneous on a 2–10 µm scale, so if areas with excess Cr are is seemingly randomly distributed (Electronic Supplement the result of Cr being concentrated in sub-100 nm inclusions Fig. S4). In addition, growth zoning is at odds with a likely of chromite, their density would need to average ca. one mantle residence of > 1 Ga for these diamonds (assuming 2 nanocrystal per 2 µm , but with a highly irregular distribu- Early Archean formation of harzburgitic diamonds and tion. More work is needed to fully understand the origin of exhumation at about 2.2 Ga; Gurney et al. 2010; Stachel Cr heterogeneity, and if it is present in other diamond-hosted and Harris 1997b), as it is inconceivable that such zon- olivine inclusion suites, particularly those where high Cr ing patterns are preserved in olivine at mantle conditions contents have been reported. (T > 1000 °C) for billions of years. Therefore, the Cr and V variability must be a relatively recent feature. One possible explanation is related to the loss of hydro- Summary and conclusions gen from olivine inclusions during diamond ascent in kim- berlitic magma. Although the unresorbed shapes of Akwatia • Diamond-hosted olivine inclusions from the Akwatia dia- diamonds indicate that residence time in the magma was mond suite were used to determine diamond paragenesis short, it has been suggested that the source of these dia- (lherzolitic vs. harzburgitic), study mantle processes such monds were komatiites (Canales and Norman 2003), which 1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 25 of 28 100

as melting and refertilisation, and determine physical that occurred before trapping of the inclusions in the dia- conditions of diamond formation (pressure, temperature, mond. oxidation state). • A proportion of Cr in olivine is present as divalent Cr, • Trace element compositions of olivine inclusions due to the low oxidation state and high temperature of can be used to determine the paragenesis of host dia- the mantle source. However, in harzburgitic olivine Cr monds. Olivines with Ca > 300 µg/g, Ca/Al > 2.2, and and V are commonly lower than expected based on the Na > 60 µg/g or Na/Al > 0.75 are lherzolitic (and co- equilibration temperatures. This is due to the absence exist with G9 or G11 garnets), whereas olivines with of cpx in harzburgitic lithologies and the resulting low Ca < 100 µg/g or Ca/Al < 2.2, Na ≤ 60 µg/g and Na/ Na activity, which reduces the uptake of Cr and V via Al ≤ 0.75 are harzburgitic (and co-exist with G10 gar- the Na–M3+ exchange mechanism, which dominates in nets). lherzolites. • To use olivine as a diamond indicator mineral, additional • Chromium, Al and V distribution in many olivines is screening is required, as some lherzolitic low-T mantle highly heterogeneous and it is recommended that multi- xenoliths may have olivine that fall into the harzburgitic ple points are analysed on each diamond-hosted olivine diamond inclusion feld. We recommend that the dataset inclusion to check for heterogeneity and to achieve a is frst fltered to olivines of mantle origin (Fo = 92–95% reliable average composition. Although the origin of the and NiO > 0.3 wt%) with Al > 40 µg/g, Ca < 300 µg/g and heterogeneity is unclear, the most likely cause is a late- Na < 60 µg/g. Then ‘G10’ olivines are those from the fl- stage process, possibly hydrogen loss during diamond tered dataset that have at least two of the following three exhumation, resulting in oxidation of ­Cr2+ to ­Cr3+ and ratios: Ca/Al < 2.2, Al/Na > 1.5 and Ca/Na < 5. If Na oversaturation of chromite. has not been measured, the fltering can be simplifed to • Olivine inclusions in diamonds can provide similar infor- Al > 40 µg/g, Ca < 300 µg/g and Ca/Al < 2.2. This results mation about equilibration temperature, diamond par- in less olivines identifed positively as ‘G10’ olivines. agenesis and mantle processes as is commonly derived • Trace-element thermometry for harzburgitic olivine is from diamond-hosted garnet inclusions. To fully utilise hampered by variation in the concentration of elements the potential of olivine in mantle studies, it is recom- which are assumed to be constant in co-existing miner- mended that at least the following suite of trace elements als, which afects Ca in particular. Aluminium appears to is analysed: Na, Al, Ca, Sc, Ti, V, Cr, Mn, Ni. be little afected, so TAl-ol estimates are equally reliable for harzburgitic as lherzolitic olivine. Ca–ol thermom- etry can only be applied to lherzolitic olivines, but even Acknowledgements We thank Richard Hinton for assistance with Ni in garnet analysis, Ben Harte for supplying Ni in garnet standards those often give anomalously high temperature estimates. and Chris Hayward for assistance with electron microprobe analysis. Trace element thermometers (TNi-grt, TAl-ol) result in simi- Constructive reviews by Peter Tollan and an anonymous reviewer are lar average temperatures as major element exchange ther- greatly appreciated. mometry but good linear correlations are not observed, indicating the possibility of disequilibrium between dif- Open Access This article is distributed under the terms of the Crea- tive Commons Attribution 4.0 International License (http://creat​iveco​ ferent silicate inclusions in the same diamond. mmons.org/licen​ ses/by/4.0/​ ), which permits unrestricted use, distribu- • Based on traditional geothermobarometry, the diamonds tion, and reproduction in any medium, provided you give appropriate of the Akwatia feld were derived from a continuous credit to the original author(s) and the source, provide a link to the mantle section from the diamond–graphite bound- Creative Commons license, and indicate if changes were made. ary (ca. 140-km depth) down to ca. 200 km along a 38–39 mW/m2 conductive geotherm. Similar results can be obtained from Al–ol thermometry, with TAl-ol ranging References from 1020 °C at 45 kbar to 1325 °C at 65 kbar, using the assumption that the olivine suite is derived from within Ballhaus C, Berry RF, Green DH (1991) High-pressure experimental the diamond stability feld at temperatures below the calibration of the olivine-ortho-pyroxene-spinel oxygen geoba- rometer—implications for the oxidation-state of the upper man- mantle adiabat. tle. Contrib Mineral Petrol 107(1):27–40. https://doi.org/10.1007/​ • Titanium in olivine correlates well with Ti in garnet, and BF003​11183​ is highest in lherzolitic olivine, whilst being extremely Ballhaus C, Berry RF, Green DH (1994) High-pressure experimental depleted (< 1 µg/g) in many harzburgitic olivines, which calibration of the olivine-ortho-pyroxene-spinel oxygen geoba- rometer—implications for the oxidation-state of the upper-mantle. indicates strong melt depletion with no cpx left in the Contrib Mineral Petrol 118(1):109. https://doi.org/10.1007/BF003​ ​ residue. Lherzolitic olivines have similar Cr# as harzbur- 10615​ gitic olivines, so their high Ti contents are indicative of Batanova VG, Sobolev AV, Kuzmin DV (2015) Trace element analysis melt refertilisation during an ancient metasomatic event of olivine: high precision analytical method for JEOL JXA-8230

1 3 100 Page 26 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

electron probe microanalyser. Chem Geol 419:149–157. https​:// Foley SF, Andronikov AV, Jacob DE, Melzer S (2006) Evidence from doi.org/10.1016/j.chemg​eo.2015.10.042 Antarctic mantle peridotite xenoliths for changes in mineralogy, Berry AJ, Shelley JMG, Foran GJ, O’Neill HS, Scott DR (2003) A geochemistry and geothermal gradients beneath a developing furnace design for XANES spectroscopy of silicate melts under rift. Geochim Cosmochim Acta 70(12):3096–3120. https​://doi. controlled oxygen fugacities and temperatures to 1773 K. J Syn- org/10.1016/j.gca.2006.03.010 chrotron Radiat 10:332–336 Foley SF, Prelevic D, Rehfeldt T, Jacob DE (2013) Minor and trace Berry AJ, Hermann J, O’Neill HSC, Foran GJ (2005) Fingerprinting elements in olivines as probes into early igneous and mantle the water site in mantle olivine. Geology 33(11):869–872. https​ melting processes. Earth Planet Sci Lett 363:181–191. https​:// ://doi.org/10.1130/g2175​9.1 doi.org/10.1016/j.epsl.2012.11.025 Berry AJ, O’Neill HSC, Hermann J, Scott DR (2007) The infrared Glaser SM, Foley SF, Gunther D (1999) Trace element compositions of signature of water associated with trivalent cations in olivine. minerals in garnet and spinel peridotite xenoliths from the Vitim Earth Planet Sci Lett 261(1–2):134–142. https://doi.org/10.1016/j.​ volcanic feld, Transbaikalia, eastern Siberia. Lithos 48(1–4):263– epsl.2007.06.021 285. https​://doi.org/10.1016/S0024​-4937(99)00032​-8 Bizimis M, Salters VJM, Bonatti E (2000) Trace and REE content of Griffin WL, Ryan CG (1996) An experimental calibration of the clinopyroxenes from supra-subduction zone peridotites: Implica- ‘‘nickel in garnet’’ geothermometer with applications—discus- tions for melting and enrichment processes in island arcs. Chem sion. Contrib Mineral Petrol 124(2):216–218 Geol 165(1–2):67–85 Grifn WL, Cousens DR, Ryan CG, Sie SH, Suter GF (1989) Ni in Brey GP, Köhler T (1990) Geothermobarometry in 4-phase lherzolites. chrome pyrope garnets—a new geothermometer. Contrib Mineral 2. New thermobarometers, and practical assessment of existing Petrol 103(2):199–202. https​://doi.org/10.1007/Bf003​78505​ thermobarometers. J Petrol 31(6):1353–1378 Grütter HS, Gurney JJ, Menzies AH, Winter F (2004) An updated Brey GP, Shu Q (2018) The birth, growth and ageing of the Kaapvaal classifcation scheme for mantle-derived garnet, for use by subcratonic mantle. Mineral Petrol 112:23–41. https​://doi. diamond explorers. Lithos 77(1–4):841–857. https​://doi. org/10.1007/s0071​0-018-0577-8 org/10.1016/j.litho​s.2004.04.012 Bulanova GP (1995) The formation of diamond. J Geochem Explor Grütter H, Latti D, Menzies A (2006) Cr-saturation arrays in concen- 53(1–3):1–23. https​://doi.org/10.1016/0375-6742(94)00016​-5 trate garnet compositions from kimberlite and their use in man- Bussweiler Y, Brey GP, Pearson DG, Stachel T, Stern RA, Hardman tle barometry. J Petrol 47(4):801–820. https​://doi.org/10.1093/ MF, Kjarsgaard BA, Jackson SE (2017) The aluminum-in-olivine petro​logy.egi09​6 thermometer for mantle peridotites—experimental versus empiri- Gurney JJ (1984) A correlation between garnets and diamonds in cal calibration and potential applications. Lithos 272:301–314. kimberlites. In: Glover JE, Harris PG (eds) Kimberlite occur- https​://doi.org/10.1016/j.litho​s.2016.12.015 rence and origin: a basis for conceptual models in exploration, Canales DG, Norman DI (2003) The Akwatia diamond feld, Ghana, vol 8. Publs Geology Department and University Extension, West Africa: source rocks. In: Geological Society of America University of Western Australia, Perth, pp 143–166 Annual Meeting, Seattle, vol 35, p 230 Gurney JJ, Helmstaedt H, Moore RO (1993) A review of the use Canil D (1996) An experimental calibration of the ‘‘nickel in garnet’’ and application of mantle mineral geochemistry in diamond geothermometer with applications—reply. Contrib Mineral Petrol exploration. Pure Appl Chem 65(12):2423–2442. https​://doi. 124(2):219–223 org/10.1351/pac19​93651​22423​ Canil D (1999a) The Ni-in-garnet geothermometer: calibration at natu- Gurney JJ, Helmstaedt HH, Richardson SH, Shirey SB (2010) Dia- ral abundances. Contrib Mineral Petrol 136(3):240–246 monds through Time. Econ Geol 105(3):689–712. https​://doi. Canil D (1999b) Vanadium partitioning between orthopyroxene, spinel org/10.2113/gseco​ngeo.105.3.689 and silicate melt and the redox states of mantle source regions for Harley SL (1984) An experimental study of the partitioning of Fe primary . Geochim Cosmochim Acta 63(3–4):557–572. and Mg between garnet and orthopyroxene. Contrib Mineral https​://doi.org/10.1016/S0016​-7037(98)00287​-7 Petrol 86(4):359–373. https​://doi.org/10.1007/bf011​87140​ Canil D (2002) Vanadium in peridotites, mantle redox and tectonic Hasterok D, Chapman DS (2011) Heat production and geotherms for environments: archean to present. Earth Planet Sci Lett 195(1– the continental lithosphere. Earth Planet Sci Lett 307(1–2):59– 2):75–90. https​://doi.org/10.1016/s0012​-821x(01)00582​-9 70. https​://doi.org/10.1016/j.epsl.2011.04.034 Canil D, Oneill HS, Pearson DG, Rudnick RL, Mcdonough WF, Hellebrand E, Snow JE, Dick HJB, Hofmann AW (2001) Coupled Carswell DA (1994) Ferric iron in peridotites and mantle oxida- major and trace elements as indicators of the extent of melting tion-states. Earth Planet Sci Lett 123(1–4):205–220. https​://doi. in mid-ocean-ridge peridotites. Nature 410:677–681 org/10.1016/0012-821x(94)90268​-2 Hermann J, O’Neill HSC, Berry AJ (2005) Titanium solubility in Chirico PG, Malpeli KC, Anum S, Phillips EC (2010) Alluvial dia- olivine in the system ­TiO2–MgO–SiO2: no evidence for an mond resource potential and production capacity assessment of ultra-deep origin of Ti-bearing olivine. Contrib Mineral Petrol Ghana. USGS Scientifc Investigations Report 2010–5045, p 25 148(6):746–760. https​://doi.org/10.1007/s0041​0-004-0637-4 Dawson JB, Stephens WE (1975) Statistical classifcation of garnets Hervig RL, Smith JV (1982) Temperature-dependent distribution from kimberlite and associated xenoliths. J Geol 83(5):589–607. of Cr between olivine and pyroxenes in lherzolite xenoliths. https​://doi.org/10.2307/30061​056 Contrib Mineral Petrol 81(3):184–189 De Hoog JCM, Gall L, Cornell DH (2010) Trace-element geochemis- Hervig RL, Smith JV, Steele IM (1980a) Fertile and barren Al-Cr- try of mantle olivine and application to mantle petrogenesis and spinel harzburgites from the upper mantle—ion and electron- geothermobarometry. Chem Geol 270(1–4):196–215. https://doi.​ probe analyses of trace-elements in olivine and ortho-pyrox- org/10.1016/j.chemg​eo.2009.11.017 ene—relation to lherzolites. Earth Planet Sci Lett 50(1):41–58 Demouchy S, Bolfan-Casanova N (2016) Distribution and transport of Hervig RL, Smith JV, Steele IM, Gurney JJ, Meyer HOA, Harris hydrogen in the lithospheric mantle: a review. Lithos 240:402– JW (1980b) Diamonds—minor elements in silicate inclu- 425. https​://doi.org/10.1016/j.litho​s.2015.11.012 sions—pressure-temperature implications. J Geophys Res Dick HJB, Bullen T (1984) Chromian spinel as a petrogenetic indica- 85(B12):6919–6929 tor in abyssal and alpine-type peridotites and spatially associated Hervig RL, Smith JV, Dawson JB (1986) Lherzolite xenoliths in Lavas. Contrib Mineral Petrol 86(1):54–76 kimberlites and : petrogenetic and crystallochemical

1 3 Contributions to Mineralogy and Petrology (2019) 174:100 Page 27 of 28 100

significance of some minor and trace elements in olivine, Matveev S, Stachel T (2007) FTIR spectroscopy of OH in olivine: a pyroxenes, garnet and spinel. Trans R Soc Edinb Earth Scis new tool in kimberlite exploration. Geochim Cosmochim Acta 77:181–201 71(22):5528–5543 Ingrin J, Skogby H (2000) Hydrogen in nominally anhy- Milman-Barris MS, Beckett JR, Baker MB, Hofmann AE, Morgan Z, drous upper-mantle minerals: concentration levels and Crowley MR, Vielzeuf D, Stolper E (2008) Zoning of phosphorus implications. Eur J Miner 12(3):543–570. https​://doi. in igneous olivine. Contrib Miner Petrol 155(6):739–765 org/10.1127/0935-1221/2000/0012-0543 Nimis P, Taylor WR (2000) Single clinopyroxene thermobarometry for Ionov DA (2010) Petrology of mantle wedge lithosphere: new data on garnet peridotites: Part I. Calibration and testing of a Cr-in-Cpx supra-subduction zone peridotite xenoliths from the Andesitic barometer and an enstatite-in-Cpx thermometer. Contrib Miner Avacha Volcano, Kamchatka. J Petrol 51(1–2):327–361. https​ Petrol 139(5):541–554 ://doi.org/10.1093/petro​logy/egp09​0 O’Neill HSC, Wood BJ (1979) Experimental study of Fe–Mg partition- Irving AJ, Frey FA (1978) Distribution of trace elements between ing between garnet and olivine and iIts calibration as a geother- garnet megacrysts and host volcanic liquids of kimber- mometer. Contrib Miner Petrol 70(1):59–70 litic to rhyolitic composition. Geochim Cosmochim Acta Papike JJ, Karner JM, Shearer CK (2005) Comparative planetary min- 42(6):771–787 eralogy: valence state partitioning of Cr, Fe, Ti, and V among Ivanic TJ (2007) The chromite-garnet peridotite assemblages and their crystallographic sites in olivine, pyroxene, and spinel from plane- role in the evolution of the mantle lithosphere. Unpublished Ph.D. tary basalts. Am Miner 90(2–3):277–290. https://doi.org/10.2138/​ thesis. University of Edinburgh, Edinburgh am.2005.1779 Jarosewich E, Nelen J, Norber J (1980) Reference samples for electron Phillips D, Harris JW, Viljoen KS (2004) Mineral chemistry and probe analysis. Geostand Newslett 4(1):43–47 thermobarometry of inclusions from De Beers Pool diamonds, Jochum KP, Willbold M, Raczek I, Stoll B, Herwig K (2005) Chemical Kimberley, South Africa. Lithos 77(1–4):155–179. https​://doi. characterisation of the USGS reference glasses GSA-1G, GSC- org/10.1016/j.litho​s.2004.04.005 1G, GSD-1G, GSE-1G, BCR-2G, BHVO-2G and BIR-1G using Pirard C, Hermann J, O’Neill HS (2013) Petrology and Geochemistry EPMA, ID-TIMS, ID-ICP-MS and LA-ICP-MS. Geostand Geoa- of the crust-mantle boundary in a Nascent Arc, Massif du Sud nal Res 29(3):285–302 , New Caledonia, SW Pacifc. J Petrol 54(9):1759–1792. Kennedy CS, Kennedy GC (1976) The equilibrium boundary between https​://doi.org/10.1093/petro​logy/egt03​0 graphite and diamond. J Geophys Res 81(14):2467–2470 Rehfeldt T, Foley SF, Jacob DE, Carlson RW, Lowry D (2008) Con- Klemme S (2004) The infuence of Cr on the garnet-spinel transition in trasting types of metasomatism in , wehrlite and web- the Earth’s mantle: experiments in the system MgO–Cr2O3–SiO2 sterite xenoliths from Kimberley, South Africa. Geochim Cos- and thermodynamic modelling. Lithos 77(1–4):639–646. https​:// mochim Acta 72(23):5722–5756. https​://doi.org/10.1016/j. doi.org/10.1016/j.litho​s.2004.03.017 gca.2008.08.020 Klemme S, Ivanic TJ, Connolly JAD, Harte B (2009) Thermodynamic Rudnick RL, Nyblade AN (1999) The thickness and heat production of modelling of Cr-bearing garnets with implications for diamond Archean lithosphere: constraints from xenolith thermobarometry inclusions and peridotite xenoliths. Lithos 112:986–991 and surface heat fow. In: Fei Y, Bertka CM, Mysen BO (eds) Köhler T, Brey GP (1988) Ca in olivine as a geobarometer for lherzo- Mantle petrology: Field observations and high pressure experi- lites. Chem Geol 70(1–2):10 mentation; a tribute to Francis R (Joe) Boyd. Special Publication Köhler TP, Brey GP (1990) Calcium exchange between olivine and No. 6, The Geochemical Society, Washington, pp 3–12 clinopyroxene calibrated as a geothermobarometer for natural Russell JK, Porritt LA, Lavallee Y, Dingwell DB (2012) Kimberlite peridotites from 2 to 60 kb with applications. Geochim Cosmo- ascent by assimilation-fuelled buoyancy. Nature 481:352–356. chim Acta 54(9):2375–2388 https​://doi.org/10.1038/natur​e1074​0 Krogh EJ (1988) The garnet-clinopyroxene Fe–Mg geothermometer—a Ryan CG, Grifn WL, Pearson NJ (1996) Garnet geotherms: pressure- reinterpretation of existing experimental-data. Contrib Mineral temperature data from Cr-pyrope garnet xenocrysts in volcanic Petrol 99(1):44–48 rocks. J Geophys Res Solid Earth 101(B3):5611–5625 Kurosawa M, Yurimoto H, Sueno S (1997) Patterns in the hydrogen Schreiber HD, Merkel RC, Schreiber VL, Balazs GB (1987) Mutual and trace element compositions of mantle olivines. Phys Chem interactions of redox couples via electron exchange in silicate Miner 24(6):385–395 melts—models for geochemical melt systems. J Geophys Res- Li JP, O’Neill HSC, Seifert F (1995) Subsolidus phase-relations in the Solid 92(B9):9233–9245. https​://doi.org/10.1029/Jb092​ib09p​ system MgO–SiO2–Cr–O in equilibrium with metallic Cr, and 09233​ their signifcance for the petrochemistry of chromium. J Petrol Shchukina EV, Shchukin VS (2018) Diamond exploration potential of 36(1):107–132 the Northern East European Platform. Minerals 8(5):189. https://​ Luth RW, Stachel T (2014) The bufering capacity of lithospheric doi.org/10.3390/min80​50189​ mantle: implications for diamond formation. Contrib Miner Petrol Smith EM, Shirey SB, Richardson SH, Nestola F, Bullocks ES, Wang 168:1083. https​://doi.org/10.1007/s0041​0-014-1083-6 JH, Wang WY (2018) Blue boron-bearing diamonds from Earth’s Mallmann G, O’Neill HSC (2009) The crystal/melt partitioning of V lower mantle. Nature 560:84–87. https​://doi.org/10.1038/s4158​ during mantle melting as a function of oxygen fugacity compared 6-018-0334-5 with some other elements (Al, P, Ca, Sc, Ti, Cr, Fe, Ga, Y, Zr Sobolev NV, Lavrente YG, Pokhilenko NP, Usova LV (1973) Chrome- and Nb). J Petrol 50(9):1765–1794. https://doi.org/10.1093/petro​ ​ rich garnets from kimberlites of Yakutia and their parageneses. logy/egp05​3 Contrib Miner Petrol 40(1):39–52. https​://doi.org/10.1007/bf003​ Mallmann G, O’Neill HSC, Klemme S (2009) Heterogeneous distribu- 71762​ tion of phosphorus in olivine from otherwise well-equilibrated Sobolev NV, Logvinova AM, Zedgenizov DA, Pokhile- spinel peridotite xenoliths and its implications for the mantle geo- nko NP, Kuzmin DV, Sobolev AV (2008) Olivine inclu- chemistry of lithium. Contrib Miner Petrol 158(4):485–504. https​ sions in Siberian diamonds: high-precision approach to ://doi.org/10.1007/s0041​0-009-0393-6 minor elements. Eur J Miner 20(3):305–315. https​://doi. Matsyuk SS, Langer K (2004) Hydroxyl in olivines from mantle xeno- org/10.1127/0935-1221/2008/0020-1829 liths in kimberlites of the Siberian platform. Contrib Miner Petrol Sobolev NV, Logvinova AM, Zedgenizov DA, Pokhilenko NP, 147(4):413–437. https​://doi.org/10.1007/s0041​0-003-0541-3 Malygina EV, Kuzmin DV, Sobolev AV (2009) Petrogenetic

1 3 100 Page 28 of 28 Contributions to Mineralogy and Petrology (2019) 174:100

signifcance of minor elements in olivines from diamonds and Stosch HG (1981) Sc, Cr, Co and Ni partitioning between miner- peridotite xenoliths from kimberlites of Yakutia. Lithos 112:701– als from spinel peridotite xenoliths. Contrib Miner Petrol 713. https​://doi.org/10.1016/j.litho​s.2009.06.038 78(2):166–174 Stachel T, Harris JW (1997a) Diamond precipitation and mantle meta- Sutton SR, Bajt AS, Rivers ML, Smith JV (1993) X-ray microprobe somatism—evidence from the trace element chemistry of silicate determination of chromium oxidation state in olivine from lunar inclusions in diamonds from Akwatia, Ghana. Contrib Miner Pet- and kimberlitic diamonds. In: 24­ th Lunar and Planetary Sci- rol 129(2–3):143–154 ence Conference. Houston, TX, pp 1383–1384 Stachel T, Harris JW (1997b) Syngenetic inclusions in diamond Tollan PME, Dale CW, Hermann J, Davidson JP, Arculus RJ (2017) from the Birim feld (Ghana)—a deep peridotitic profle with Generation and modifcation of the mantle wedge and lithosphere a history of depletion and re-enrichment. Contrib Miner Petrol beneath the west bismarck island arc: melting, metasomatism and 127(4):336–352 thermal history of peridotite xenoliths from Ritter Island. J Petrol Stachel T, Harris JW (2008) The origin of cratonic diamonds—con- 58(8):1475–1510. https​://doi.org/10.1093/petro​logy/egx06​2 straints from mineral inclusions. Ore Geol Rev 34(1–2):5–32 Tollan PME, O’Neill HS, Hermann J (2018) The role of trace ele- Stachel T, Viljoen KS, Brey G, Harris JW (1998) Metasomatic pro- ments in controlling H incorporation in San Carlos olivine. Con- cesses in lherzolitic and harzburgitic domains of diamondiferous trib Miner Petrol 173(11):ARTN89 lithospheric mantle: REE in garnets from xenoliths and inclusions Witt-Eickschen G, O’Neill HS (2005) The efect of temperature on in diamonds. Earth Planet Sci Lett 159(1–2):1–12. https​://doi. the equilibrium distribution of trace elements between clinopy- org/10.1016/S0012​-821X(98)00064​-8 roxene, orthopyroxene, olivine and spinel in upper mantle peri- Stagno V, Ojwang DO, McCammon CA, Frost DJ (2013) The oxida- dotite. Chem Geol 221(1–2):65–101. https​://doi.org/10.1016/j. tion state of the mantle and the extraction of carbon from Earth’s chemg​eo.2005.04.005 interior. Nature 493:84–88. https​://doi.org/10.1038/natur​e1167​9 Wu CM, Zhao GC (2007) A recalibration of the garnet-olivine geo- Stead CV, Tomlinson EL, Kamber BS, Babechuk MG, McKenna CA thermometer and a new geobarometer for garnet peridotites and (2017) Rare earth element determination in olivine by laser abla- garnet-olivine-plagioclase-bearing granulites. J Metamorph Geol tion-quadrupole-ICP-MS: an analytical strategy and applications. 25(5):497–505 Geostand Geoanal Res 41(2):197–212. https​://doi.org/10.1111/ ggr.12157​ Publisher’s Note Springer Nature remains neutral with regard to Steele IM, Hervig RL, Hutcheon ID, Smith JV (1981) Ion microprobe jurisdictional claims in published maps and institutional afliations. techniques and analyses of olivine and low-Ca pyroxene. Am Miner 66:526–546

1 3